This is the accessible text file for GAO report number GAO-11-853 
entitled 'Freight Railroad Safety: Hours of Service Changes Have 
Increased Rest Time, but More Can Be Done to Address Fatigue Risks' 
which was released on October 28, 2011. 

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as 
part of a longer term project to improve GAO products' accessibility. 
Every attempt has been made to maintain the structural and data 
integrity of the original printed product. Accessibility features, 
such as text descriptions of tables, consecutively numbered footnotes 
placed at the end of the file, and the text of agency comment letters, 
are provided but may not exactly duplicate the presentation or format 
of the printed version. The portable document format (PDF) file is an 
exact electronic replica of the printed version. We welcome your 
feedback. Please E-mail your comments regarding the contents or 
accessibility features of this document to Webmaster@gao.gov. 

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. Because this work 
may contain copyrighted images or other material, permission from the 
copyright holder may be necessary if you wish to reproduce this 
material separately. 

United States Government Accountability Office: 
GAO: 

Report to Congressional Requesters: 

September 2011: 

Freight Railroad Safety: 

Hours of Service Changes Have Increased Rest Time, but More Can Be 
Done to Address Fatigue Risks: 

GAO-11-853: 

GAO Highlights: 

Highlights of GAO-11-853, a report to congressional requesters. 

Why GAO Did This Study: 

The Rail Safety Improvement Act of 2008 (RSIA) overhauled requirements 
for how much time certain freight railroad workers can spend on the 
job (called “hours of service”). Changes included limiting the number 
of consecutive days on duty before rest is required, increasing 
minimum rest time from 8 to 10 hours, and requiring rest time to be 
undisturbed. RSIA also provided for pilot projects and waivers. 
RSIA’s changes became effective for freight railroads in July 2009. 
GAO was asked to assess (1) the impact of these changes on covered 
train and engine (T&E) employees, including implications for fatigue, 
(2) the impact of the changes on the rail industry, and (3) actions 
the Federal Railroad Administration (FRA) has taken to oversee 
compliance with hours of service requirements and implement RSIA 
provisions for pilot projects and waivers. To perform this work, GAO 
analyzed covered employee work schedules and used models to assess 
fatigue, surveyed the railroad industry, analyzed FRA inspection and 
enforcement data, and interviewed federal and railroad officials as 
well as fatigue and sleep experts. 

What GAO Found: 

According to GAO’s analysis of covered employee work schedules, RSIA’s 
requirements led to changed work schedules, increased rest time, and 
reduced risk of fatigue for covered T&E employees. RSIA's consecutive 
work day limits and rest requirements contributed to work schedule 
changes and increases in rest time. Increased rest time also led to 
equivalent decreases in the hours that covered employees worked. 
Overall, GAO found, using an FRA-validated fatigue model, that the 
time covered employees spent working at a high risk of fatigue—a level 
associated with reduced alertness and an increased risk of errors and 
accidents—decreased by about 29 percent for employees of class I 
railroads (those with the largest revenues) and by about 36 percent 
for employees of selected class II railroads (those with smaller 
revenues). GAO’s analysis also shows that there are further 
opportunities to reduce fatigue risk. Specifically, RSIA's changes did 
not result in material decreases in night work, yet scientific 
literature and GAO’s analysis show night work represents a major 
factor in fatigue risk. 

As might be expected from changes aimed at improving safety by 
reducing covered employee fatigue, the railroad industry reported that 
RSIA’s hours of service changes had operational and administrative 
effects on it, some of which increased some railroads’ one-time or 
ongoing costs. GAO did not determine how RSIA’s changes affected 
railroads’ earnings; but the act took effect as the economy was 
starting to recover from the recession that began in late 2008. 
Through its industry survey and interviews, GAO found that RSIA’s 
changes affected railroad operations, including changes to crew and 
train schedules and increases in staffing levels. Railroad officials 
GAO spoke with attributed these changes to RSIA’s consecutive work day 
limits and rest requirements, both of which acted to reduce people’s 
availability to work. To maintain operations while complying with the 
law, railroad officials told GAO they, among other things, hired new 
employees or brought employees back from furlough. GAO estimated that 
adding people—120 to 500 each by some class I railroads—increased 
these railroads’ annual costs by $11 million to $50 million. 
Administrative effects reported by railroads included a need for 
railroads to revise their hours of service timekeeping systems. 

FRA uses a risk-based approach to oversee compliance with hours of 
service and other safety requirements, analyzing inspection and 
accident data to help target inspections to activities where 
noncompliance is associated with a greater risk of accidents. GAO’s 
analysis of inspection and enforcement data for the years before RSIA 
took effect and for the following year show it is too early to 
determine if FRA has changed the priority it assigns to overseeing 
hours of service requirements or if a change in priority is warranted. 
FRA has not been able to implement RSIA-required pilot projects 
because no railroads have chosen to participate. Nor has it approved 
voluntary pilot projects designed to test the fatigue-reduction 
potential of alternatives to RSIA requirements. FRA has approved 
petitions for waivers of compliance with hours of service requirements 
for some railroads, but is not required by RSIA to collect data on the 
safety effects of the approved alternatives. Data from pilot projects-—
if implemented—-and waivers could be used to improve FRA’s assessment 
of fatigue issues. 

What GAO Recommends: 

FRA should, among other things, assess the fatigue risk of work 
performed during night hours and develop data from pilot projects and 
waivers to help assess fatigue issues. The Department of 
Transportation raised concerns about findings related to the oversight 
process and provided additional clarifying information. Based in part 
on this additional information, GAO withdrew part of a recommendation. 
GAO also made other clarifications in the report. 

View [hyperlink, http://www.gao.gov/products/GAO-11-853] or key 
components. To view the e-supplement online, click [hyperlink, 
http://www.gao.gov/products/GAO-11-894SP]. For more information, 
contact Susan Fleming at (202) 512-2834 or flemings@gao.gov. 

[End of section] 

Contents: 

Letter: 

Background: 

RSIA's Requirements Increased Rest Time and Decreased Fatigue Risk 
While Leaving Opportunities for Further Reductions in Fatigue Risk 
from Night Work: 

New Hours of Service Requirements Have Led to Operational and 
Administrative Changes That Have Increased Some Railroads' Costs: 

Too Soon to Assess FRA's Hours of Service Oversight, but 
Implementation of Pilot Project and Waiver Provisions Could Yield More 
Data to Address Fatigue: 

Conclusions: 

Recommendations for Executive Action: 

Agency Comments and Our Evaluation: 

Appendix I: Objectives, Scope, and Methodology: 

Appendix II: The Development of Biomathematical Models of Fatigue: 

Appendix III: GAO Contact and Staff Acknowledgments: 

Tables: 

Table 1: Selected Hours of Service Requirements for Freight Railroad 
Train Employees: 

Table 2: Hours of Service and Hours of Service Recordkeeping 
Inspections of Freight Railroads, Fiscal Years 2005-2010: 

Table 3: Hours of Service and Hours of Service Recordkeeping 
Enforcement Actions Taken against Freight Railroads, Fiscal Years 2005-
2010: 

Table 4: Disposition of Petitions for Waivers of Limitations on Hours 
of Service for Railroad Employees, May 2009 through June 2011: 

Figures: 

Figure 1: Train Accidents per Million Train-miles, by Class of 
Railroad, 2001-2010: 

Figure 2: Major Causes of Train Accidents, 2010: 

Figure 3: Human-Factor-Caused Train Accidents per Million Train-miles, 
by Class of Railroad, 2001-2010: 

Figure 4: Class I and Class II Railroad High Fatigue Risk Level 
Changes for Covered T&E Employees Based on Total Hours Worked, May 
2008 and May 2010: 

Figure 5: Differences in Fatigue Risk for Hypothetical Daytime and 
Nighttime Work Schedule: 

Figure 6: Degree of Burden on Surveyed Railroads from Changing Train 
Schedules As a Direct Result of RSIA's Hours of Service Changes: 

Figure 7: Number of Surveyed Railroads That Did Not Meet Guaranteed 
Hours As a Direct Result of RSIA's Requirements, by Class of Railroad: 

Figure 8: Railroad Industry Responses about Effects on Customer 
Service As a Direct Result of RSIA Hours of Service Changes, by Class 
of Railroad: 

Figure 9: Changes Railroads, or Their Parent Companies, Made Related 
to Timekeeping Systems As a Direct Result of RSIA's Hours of Service 
Changes, by Class of Railroad: 

Figure 10: Proportion of Hours of Service and Hours of Service 
Recordkeeping Inspections Conducted by Selected FRA Disciplines, and 
As a Portion of All Operating Practices, Fiscal Years 2005-2010: 

Figure 11: Hours of Service and Hours of Service Recordkeeping 
Inspections of Freight Railroads, Fiscal Years 2005-2010: 

View GAO-11-894SP key component: 

Freight Railroad Safety: Results of Rail Industry Survey about Hours 
of Service Issues [hyperlink, 
http://www.gao.gov/products/GAO-11-894SP], September 2011, an E-
supplement to GAO-11-853: 

Abbreviations List: 

ASLRRA: American Short Line and Regional Railroad Association: 

DOT: Department of Transportation: 

FAID: Fatigue Audit InterDyneTM: 

FAST: Fatigue Avoidance Scheduling ToolTM: 

FRA: Federal Railroad Administration: 

NIP: National Inspection Plan: 

RSIA: Rail Safety Improvement Act of 2008: 

T&E: train and engine: 

[End of section] 

United States Government Accountability Office: 
Washington, DC 20548: 

[End of section] 

September 28, 2011: 

The Honorable John D. Rockefeller, IV: 
Chairman: 
The Honorable Kay Bailey Hutchison: 
Ranking Member: 
Committee on Commerce, Science, and Transportation: 
United States Senate: 

The Honorable Frank R. Lautenberg: 
Chairman: 
Subcommittee on Surface Transportation and Merchant Marine 
Infrastructure, Safety, and Security: 
Committee on Commerce, Science, and Transportation: 
United States Senate: 

The Honorable John Thune: 
United States Senate: 

Railroads make important contributions to the nation's economy. 
Currently, they account for about 40 percent of all freight carried 
(as measured by ton-miles[Footnote 1]), and, according to the 
Department of Transportation (DOT), the demand for freight rail 
service is expected to increase 88 percent by 2035. Although railroad 
safety has improved in recent years--as evidenced by a nearly 40-
percent decline in accidents from 2001 to 2010--accidents can have 
significant consequences, causing deaths, injuries, and property 
damage. Fatigue can be a factor in accidents, as in July 2005, when 
two freight trains collided head-on in Mississippi, resulting in the 
deaths of 4 crew members and over $9.5 million in property damages. 

To improve railroad safety by reducing the potential for fatigue, laws 
dating as far back as 1907 have limited the amount of time that 
railroads may require or allow certain railroad employees to remain on 
duty (called "hours of service") and have established minimum rest 
times between work shifts.[Footnote 2] In October 2008, Congress 
passed the Rail Safety Improvement Act of 2008 (RSIA), which called 
for an extensive overhaul of railroad safety requirements, including 
hours of service requirements.[Footnote 3] RSIA generally retained the 
maximum time train and signal employees may remain on duty (12 hours), 
but increased the minimum rest time between work shifts from 8 hours 
to 10 hours and limited the number of consecutive days that train 
employees may work before having to take a minimum rest period, among 
other requirements.[Footnote 4] RSIA also required the Secretary of 
Transportation (the Secretary) to conduct pilot projects to analyze 
practices that may be used to reduce fatigue for train and engine 
(T&E) employees and allowed for the approval of waivers of one of the 
new hours of service requirements, or for other requirements as part 
of an approved pilot project. This authority has been delegated to the 
Federal Railroad Administrator (FRA).[Footnote 5] The new hours of 
service requirements became effective for freight railroads in July 
2009. 

You asked us to review the safety and other impacts of the new hours 
of service requirements that were established in RSIA and the federal 
government's role in monitoring and enforcing the new rules. This 
report discusses the (1) impacts of the hours of service changes on 
the covered T&E workforce, including potential impacts on fatigue, (2) 
operational and administrative impacts of the hours of service changes 
on the railroad industry, and (3) actions taken by FRA to oversee 
compliance with hours of service requirements and implement RSIA 
provisions related to hours of service pilot projects and waivers. At 
the time of our work, hours of service requirements were still being 
developed for train employees in the commuter and intercity passenger 
rail industry, which DOT issued in August 2011.[Footnote 6] 
Consequently, this report focuses on freight railroad hours of service 
issues. 

To address the impacts of the hours of service changes on the covered 
T&E workforce, including potential impacts on fatigue, we reviewed 
literature related to work and rest schedules and their relationship 
to fatigue and interviewed experts who had conducted research on 
fatigue and work. We then analyzed work schedule data for covered T&E 
employees for May 2008 and May 2010. We selected these months to 
represent time periods before and after RSIA's implementation and to 
avoid the time period during which the economic recession that began 
in late 2008 was causing a rapid contraction in the rail industry. We 
discussed the time frames for our analysis with railroad officials, 
and they generally agreed with our selection. Our analysis covered 
work schedules for T&E employees at all 7 class I railroads and 6 
class II railroads that use electronic hours of service recordkeeping 
systems.[Footnote 7] To assist in this analysis, we acquired two 
models that FRA has validated for use in assessing potential fatigue 
levels of covered railroad workers--the Fatigue Audit 
InterDyneTM(FAID) model and the Fatigue Avoidance Scheduling 
ToolTTM(FAST). To help analyze RSIA's effects on the covered workforce 
at class III railroads, we conducted focused interviews with 69 class 
III railroads. To address the operational and administrative impacts 
of RSIA's hours of service changes on the railroad industry, we 
surveyed all class I and class II railroads and 232 class III 
railroads. We received responses from all 7 class I railroads, 14 of 
15 class II railroads, and 163 of 232 class III railroads, for an 
overall response rate of 72 percent. We also interviewed federal, 
railroad, and railroad trade association officials, as well as rail 
labor representatives. Unless otherwise specified, all statistics or 
coefficient estimates mentioned throughout this report are 
significantly different from zero with probability of error less than 
0.05.[Footnote 8] To address actions taken by FRA to oversee hours of 
service compliance and to implement RSIA provisions related to pilot 
projects and waivers, we reviewed legislation, regulations, and policy 
documents and analyzed FRA data on hours of service inspection and 
enforcement actions. We also reviewed information on FRA's actions to 
implement pilot projects related to hours of service issues and 
analyzed data on petitions for waivers of hours of service 
requirements and their status. Finally, we discussed inspection and 
enforcement issues with officials in FRA headquarters and regional 
offices, and with railroad officials from class I, II, and III 
railroads. (For more information on our objectives, scope, and 
methodology, see appendix I, and for the results of our railroad 
industry survey, see GAO-11-894SP.) 

We conducted this performance audit from April 2010 to September 2011 
in accordance with generally accepted government auditing standards. 
Those standards require that we plan and perform the audit to obtain 
sufficient, appropriate evidence to provide a reasonable basis for our 
findings and conclusions based on our audit objectives. We believe 
that the evidence obtained provides a reasonable basis for our 
findings and conclusions based on our audit objectives. 

Background: 

Freight railroads are an important component of the nation's 
transportation system, operating over 700 million train-miles in 2010. 
[Footnote 9] The freight railroad industry is primarily composed of 7 
large railroads (called class I railroads) and about 570 smaller class 
II and III railroads. Within the industry, class I railroads 
predominate, representing about 93 percent of total freight revenue 
and about 68 percent of total rail mileage operated in the United 
States in 2009. Class II and III railroads include regional and short 
line railroads. Regional railroads typically operate 400 to 650 miles 
of track spanning several states, while short line railroads typically 
perform point-to-point service over short distances. According to the 
American Short Line and Regional Railroad Association (ASLRRA), the 
average length of short line service is 90 miles, and over 58 percent 
of short line carriers connect with more than 1 class I railroad. As 
the association points out, short line railroads generally operate the 
first mile and last mile of U.S. freight rail commerce. 

Because railroads operate across millions of train-miles every year, 
safety is an important concern. In general, railroad safety has 
improved over the last 10 years. For example, the approximately 1,800 
freight train accidents reported to FRA in 2010 represents a decrease 
of nearly 40 percent from the approximately 3,000 train accidents 
reported in 2001.[Footnote 10] Similarly, the number of accidents per 
million train-miles for all railroads reported to FRA decreased to 2.6 
in 2010 from 4.2 in 2001 (see figure 1). Yet this decline is not equal 
for railroads of all sizes: In 2010, the rate reported for class III 
railroads, 7.1 accidents per million train-miles, was more than twice 
the rate reported for all railroads. FRA attributed the difference in 
accident rates to differences in operations between larger railroads 
(which generally operate over longer distances and perform little 
switching) and smaller ones (which generally operate over shorter 
distances and perform frequent switching).[Footnote 11] Because RSIA 
was passed by Congress in late 2008, FRA officials told us it is too 
early to tell what effect, if any, requirements contained in the law 
may have had on railroad accident rates. They said it may well take 
years to identify any particular effects.[Footnote 12] 

Figure 1: Train Accidents per Million Train-miles, by Class of 
Railroad, 2001-2010: 

[Refer to PDF for image: multiple line graph] 

Accidents (per million train-miles): 

Year: 2001; 
All railroads: 4.2; 
Class 1 (excluding Amtrak): 4.1; 
Class 2 (more than 400K Hours, not Class I): 4.2; 
Class 3 (less than 400K Hours, not Class I): 14.6. 

Year: 2002; 
All railroads: 3.8; 
Class 1 (excluding Amtrak): 3.7; 
Class 2 (more than 400K Hours, not Class I): 4.3; 
Class 3 (less than 400K Hours, not Class I): 11.2 . 

Year: 2003; 
All railroads: 4.1; 
Class 1 (excluding Amtrak): 4; 
Class 2 (more than 400K Hours, not Class I): 4.6; 
Class 3 (less than 400K Hours, not Class I): 11.8. 

Year: 2004; 
All railroads: 4.4; 
Class 1 (excluding Amtrak): 4.4; 
Class 2 (more than 400K Hours, not Class I): 4.5; 
Class 3 (less than 400K Hours, not Class I): 13.6. 

Year: 2005; 
All railroads: 4.1; 
Class 1 (excluding Amtrak): 4.1; 
Class 2 (more than 400K Hours, not Class I): 4.1; 
Class 3 (less than 400K Hours, not Class I): 12.2. 

Year: 2006; 
All railroads: 3.7; 
Class 1 (excluding Amtrak): 3.6; 
Class 2 (more than 400K Hours, not Class I): 3.9; 
Class 3 (less than 400K Hours, not Class I): 12.6. 

Year: 2007; 
All railroads: 3.4; 
Class 1 (excluding Amtrak): 3.3; 
Class 2 (more than 400K Hours, not Class I): 3.5; 
Class 3 (less than 400K Hours, not Class I): 11.9. 

Year: 2008; 
All railroads: 3.2; 
Class 1 (excluding Amtrak): 3.1; 
Class 2 (more than 400K Hours, not Class I): 3.1; 
Class 3 (less than 400K Hours, not Class I): 9.3. 

Year: 2009; 
All railroads: 2.9; 
Class 1 (excluding Amtrak): 2.8; 
Class 2 (more than 400K Hours, not Class I): 3; 
Class 3 (less than 400K Hours, not Class I): 8.5. 

Year: 2010; 
All railroads: 2.6; 
Class 1 (excluding Amtrak): 2.5; 
Class 2 (more than 400K Hours, not Class I): 3.1; 
Class 3 (less than 400K Hours, not Class I): 7.1. 

Source: GAO analysis of FRA data. 

Note: For this analysis, we used FRA's division of the railroad 
industry based on total annual work hours instead of total annual 
operating revenues. Class I railroads are the same under both 
divisions. Class II railroads report 400,000 total annual work hours 
or more but are not class I railroads, and class III railroads report 
less than 400,000 total annual work hours. 

[End of figure] 

Train accidents can be caused by a variety of factors. In 2010, human 
factors--which include such things as failure to use brakes, 
impairment of an employee's physical condition, and failure to comply 
with signals--was second only to track (e.g., broken or worn rail) as 
the primary cause of accidents (see figure 2).[Footnote 13] As for 
train accidents overall, the rate for accidents caused by human 
factors has generally decreased over the last 10 years, and again, the 
rate is higher for class III railroads than for either class I or 
class II railroads (see figure 3). Although there is a general 
downward trend, FRA attributed the decrease since 2008 to changes it 
made that year to certain safety regulations to increase railroads' 
accountability for implementing and complying with sound operating 
procedures.[Footnote 14] For example, as of January 1, 2009, every 
railroad was to have a written program of operational tests and 
inspections in effect, and the programs were to emphasize those 
operating rules that cause or are likely to cause the most accidents 
and incidents. 

Figure 2: Major Causes of Train Accidents, 2010: 

[Refer to PDF for image: pie-chart] 

Track caused: 35%; 
Human factor caused: 33%; 
Miscellaneous caused: 15%; 
Motive power/equipment caused: 13%; 
Signal caused (all track types): 4%. 

Source: GAO analysis of FRA data. 

[End of figure] 

Figure 3: Human-Factor-Caused Train Accidents per Million Train-miles, 
by Class of Railroad, 2001-2010: 

[Refer to PDF for image: multiple line graph] 

Human factor accidents (per million train-miles): 

Year: 2001; 
All Railroads: 1.45; 
Class 1 (excluding Amtrak): 1.47; 
Class 2 (less than 400K Hours, Not Class I): 1.71; 
Class 3 (more than 400K Hours, Not Group 2 or Class I): 3.17. 

Year: 2002; 
All Railroads: 1.44; 
Class 1 (excluding Amtrak): 1.45; 
Class 2 (less than 400K Hours, Not Class I): 1.74; 
Class 3 (more than 400K Hours, Not Group 2 or Class I): 3.23. 

Year: 2003; 
All Railroads: 1.65; 
Class 1 (excluding Amtrak): 1.72; 
Class 2 (less than 400K Hours, Not Class I): 1.87; 
Class 3 (more than 400K Hours, Not Group 2 or Class I): 3.26. 

Year: 2004; 
All Railroads: 1.75; 
Class 1 (excluding Amtrak): 1.78; 
Class 2 (less than 400K Hours, Not Class I): 2.21; 
Class 3 (more than 400K Hours, Not Group 2 or Class I): 3.87. 

Year: 2005; 
All Railroads: 1.61; 
Class 1 (excluding Amtrak): 1.65; 
Class 2 (less than 400K Hours, Not Class I): 1.62; 
Class 3 (more than 400K Hours, Not Group 2 or Class I): 3.57. 

Year: 2006; 
All Railroads: 1.32; 
Class 1 (excluding Amtrak): 1.31; 
Class 2 (less than 400K Hours, Not Class I): 1.54; 
Class 3 (more than 400K Hours, Not Group 2 or Class I): 3.25. 

Year: 2007; 
All Railroads: 1.32; 
Class 1 (excluding Amtrak): 1.33; 
Class 2 (less than 400K Hours, Not Class I): 1.42; 
Class 3 (more than 400K Hours, Not Group 2 or Class I): 3.67. 

Year: 2008; 
All Railroads: 1.18; 
Class 1 (excluding Amtrak): 1.17; 
Class 2 (less than 400K Hours, Not Class I): 1.37; 
Class 3 (more than 400K Hours, Not Group 2 or Class I): 2.86. 

Year: 2009; 
All Railroads: 0.98; 
Class 1 (excluding Amtrak): 0.98; 
Class 2 (less than 400K Hours, Not Class I): 1.11; 
Class 3 (more than 400K Hours, Not Group 2 or Class I): 2.38. 

Year: 2010; 
All Railroads: 0.86; 
Class 1 (excluding Amtrak): 0.89; 
Class 2 (less than 400K Hours, Not Class I): 1.04; 
Class 3 (more than 400K Hours, Not Group 2 or Class I): 1.87. 

Source: GAO analysis of FRA data. 

Note: For this analysis, we used FRA's division of the railroad 
industry based on total annual work hours instead of STB's division 
based primarily total annual operating revenues. Class I railroads are 
the same under both divisions. Class II railroads report 400,000 total 
annual work hours or more but are not class I railroads, and class III 
railroads report less than 400,000 total annual work hours. 

[End of figure] 

Beginning with the Hours of Service Act of 1907, hours of service 
requirements for certain railroads have been governed by statute. 
[Footnote 15] The 1907 act limited the work shifts of employees 
involved in train movement to 16 hours. Amendments to this law in 1969 
reduced the maximum time on duty for train employees from 16 to 14 
hours effective immediately, with a further reduction to 12 hours 
automatically taking effect 2 years later.[Footnote 16] Amendments in 
1976 established hours of service requirements for signal 
employees.[Footnote 17] RSIA represented an extensive overhaul of the 
hours of service law for train employees. Among other things, it 
limited the number of consecutive days on duty without a required rest 
period, increased minimum rest periods, required those rest periods to 
be undisturbed, and placed limitations (caps) on the cumulative total 
work hours each month (see table 1). RSIA however, allows for the 
Secretary to waive, under certain conditions, compliance with one 
provision of the hours of service requirements, namely the consecutive 
day work limits.[Footnote 18] RSIA also required the Secretary to 
conduct two pilot projects no later than October 2010 to analyze 
specific practices which may be used to reduce fatigue. Finally, RSIA 
authorized railroads and nonprofit employee labor organizations 
representing covered service employees to jointly petition the 
Secretary for approval to establish a pilot project to demonstrate 
potential benefits of implementing alternatives to strict application 
of these requirements.[Footnote 19] As shown in the table, RSIA did 
not include any specific rules for nighttime operations. In addition 
to strengthening hours of service requirements, RSIA required the 
Secretary to develop a long-term strategy for improving railroad 
safety to cover a period of not less than 5 years, certain railroads 
to develop and submit to the Secretary plans to implement positive 
train control[Footnote 20] by December 2015, and certain railroads to 
develop safety risk reduction programs, including a fatigue management 
program. In addition, under RSIA, the Secretary is to promulgate rules 
on a variety of safety issues. These responsibilities have been 
delegated to FRA. 

Table 1: Selected Hours of Service Requirements for Freight Railroad 
Train Employees: 

Description: Limitations on time on duty in a single tour; 
Pre-RSIA requirement: 12 consecutive hours of time on duty or 12 
nonconsecutive hours on duty if broken by an interim release of at 
least 4 consecutive hours, in a 24-hour period that begins at the 
beginning of the duty tour; 
Post-RSIA requirement: 12 consecutive hours of time on duty or 12 
nonconsecutive hours on duty if broken by an interim release of at 
least 4 consecutive hours uninterrupted by communication from the 
railroad likely to disturb rest, in a 24-hour period that begins at 
the beginning of the duty tour. 

Description: Limitations on consecutive duty tours; 
Pre-RSIA requirement: n/a; 
Post-RSIA requirement: May not be on duty as a train employee after 
initiating an on-duty period on 6 consecutive days without 48 
consecutive hours off duty free from any service for any rail carrier 
at the employee's home terminal.[B] Employees are permitted to 
initiate a 7th consecutive day when the employee ends the 6th 
consecutive day at the away-from-home terminal,b as part of a pilot 
project, or as part of a grandfathered collectively bargained 
arrangement. 

Description: Contact during rest time; 
Pre-RSIA requirement: Railroads permitted to communicate with covered 
employees during rest time though some communications may be 
considered service for the railroad; 
Post-RSIA requirement: A railroad may not communicate with covered 
employees during the statutory minimum off-duty period of 10 
consecutive hours, except in cases of emergency. If an employee's rest 
is disturbed, then the statutory minimum off-duty period begins again 
from the point of interruption. 

Description: Cumulative limits on time on duty (including commingled 
service)[A]; 
Pre-RSIA requirement: n/a; 
Post-RSIA requirement: Limited to 276 hours of time on duty, in 
deadhead transportation to a point of final release,[C] or any other 
mandatory activity for the railroad carrier during a calendar month. 

Description: Mandatory off-duty periods; 
Pre-RSIA requirement: 8 consecutive hours (10 consecutive hours if 
time on duty reaches 12 consecutive hours); 
Post-RSIA requirement: 10 consecutive hours of time off duty free from 
communication from the railroad likely to disturb rest, with 
additional time off duty if on-duty time plus time in or awaiting 
deadhead transportation to final release exceeds 12 hours; 
48 consecutive hours off duty, free from any service for any railroad 
carrier, after initiating an on-duty period for 6 consecutive days. 
Covered employees may initiate a 7th consecutive day of service if the 
end of a 6th day of service was at an away-from-home terminal[B] as 
part of a pilot project, or as part of a grandfathered collectively 
bargained arrangement. If 7 consecutive days are permitted, mandatory 
off-duty period extended to 72 consecutive hours. 

Description: Cumulative limits on time on duty (limbo)[C]; 
Pre-RSIA requirement: n/a; 
Post-RSIA requirement: An employee may not spend more than 30 hours 
per calendar month on duty and waiting for or in deadhead[C] 
transportation to a point of final release after reaching 12 
consecutive hours of time on duty and waiting for or being in deadhead 
transportation to a point of final release. 

Description: Special rules for nighttime operations; 
Pre-RSIA requirement: n/a; 
Post-RSIA requirement: n/a. 

Source: GAO analysis of RSIA and FRA documents. 

Note: N/A means not applicable. 

[A] Commingled service for T&E and signal employees is any noncovered 
service performed at the behest of the railroad and for the railroad 
that is not separated from covered service by a qualifying statutory 
off-duty period of 8 or 10 hours or more. For dispatchers, it is any 
noncovered service mandated by the railroad and performed for the 
railroad within any 24-hour period containing covered service. 
Commingled service is counted as time on duty. 

[B] Home terminal is the term for the geographic area containing the 
point where the employee generally reports for duty. An away-from-home 
terminal is a term for a geographic area where an employee ends a 
shift, other than a home terminal, and takes his or her statutory 
minimum off-duty period before continuing on to either a home or an 
away-from-home terminal on their next shift. 

[C] Limbo time means a period of time treated as neither time on duty 
nor time off duty, and any other period of service for the railroad 
that does not qualify as either covered service or commingled service. 
Deadheading is the physical relocation of a train employee from one 
point to another as a result of a railroad-issued verbal or written 
directive. Time spent in deadhead transportation from a duty 
assignment to the point of final release is limbo time. The limitation 
on limbo time was initially 40 hours per employee per month from July 
16, 2009, until October 15, 2009, with that number decreasing to 30 
hours per employee per month beginning October 16, 2009, except in 
certain situations. 

[End of table] 

Individual railroads are primarily responsible for their own safe 
operation. However, FRA is the primary federal agency responsible for 
formulating railroad safety policies and regulations and for 
monitoring and enforcing railroads' compliance with hours of service 
and other requirements. FRA has issued statutory interpretations 
related to covered freight railroad employees' duty and rest time, as 
well as regulations governing hours of service recordkeeping. FRA has 
also adopted what it views as a data-driven, risk-based approach to 
monitoring and enforcement.[Footnote 21] Under the National Rail 
Safety Action plan, implemented between 2005 and 2008, FRA used 
accident, incident, and other safety data to establish a framework to 
direct its regulatory and compliance efforts at the highest priority 
risks.[Footnote 22] The plan outlines a number of initiatives aimed at 
reducing the main types of train accidents, including those caused by 
human factors or track defects. One of these initiatives, the National 
Inspection Plan (NIP), uses accident and inspection data[Footnote 23] 
to focus inspections on areas that, according to the data, are likely 
to have safety problems before a serious accident occurs. The NIP 
provides guidance to each of FRA's eight regional offices on how its 
inspection resources should be allocated. Additionally, the Office of 
Railroad Safety issues the National Safety Program Plan, which 
provides a means of planning special-emphasis activities, such as 
inspection activities and initiatives that cross regional boundaries 
and are directed at issues of concern for railroads operating in 
multiple regions. 

To provide oversight, FRA conducts periodic inspections and takes 
enforcement action. 

* FRA inspections address five areas, called disciplines--operating 
practices, track, hazardous materials, signal and train control, and 
motive power and equipment (such as locomotives and freight rail 
cars). Each inspection discipline includes a number of activities 
related to specific requirements. For example, inspectors in the 
operating practices discipline--who perform about 80 percent of hours 
of service inspections--assess railroads' compliance with hours of 
service requirements for train and dispatching service 
employees.[Footnote 24] Typically, inspections are conducted at 
railroads' operating sites. For example, inspections of hours of 
service recordkeeping and inspections for compliance with hours of 
service limitations take place at duty stations or facilities where 
records are maintained. 

* To take enforcement action, FRA inspectors may cite violations and 
recommend assessment of civil penalties.[Footnote 25] FRA's 
enforcement policy, which is designed to concentrate enforcement 
efforts on the areas with the greatest potential safety benefits, 
specifies that before assessing penalties, inspectors should consider 
the seriousness of the condition or act, the potential safety hazards, 
and the current level of compliance of the railroad, among other 
things. FRA has statutory authority to assess civil penalties in the 
range of $650 (minimum) to $25,000 (ordinary maximum) for ordinary 
violations of its regulations. FRA may assess a penalty at the 
statutory aggravated maximum penalty of $100,000 "when a grossly 
negligent violation or a pattern of repeated violations has caused an 
imminent hazard of death or injury to individuals, or has caused death 
or injury."[Footnote 26] 

In addition to these activities, FRA conducts other types of safety 
oversight aimed at reducing train accidents, such as monitoring 
railroad safety data, investigating accidents, and reviewing and 
investigating complaints, as well as providing training for small 
railroads. Furthermore, FRA funds research and development to support 
its safety oversight, by, for example, assisting in the development of 
new regulations and revision of existing regulations. FRA also has 
authority to review and approve petitions for waivers of compliance 
with safety requirements, including exemptions from the hours of 
service laws for railroads with 15 or fewer covered service employees 
and waivers of one requirement of the hours of service law, the 
consecutive day work limits.[Footnote 27] Finally, FRA is authorized 
to approve pilot projects that may be conducted to demonstrate the 
potential safety benefits of alternatives to current safety 
requirements. 

As of July 2011, FRA had 592 rail safety positions, including about 
400 inspectors. In addition, about 170 state inspectors work with FRA 
as part of the State Rail Safety Participation Program.[Footnote 28] 
As of 2009, the railroad industry had about 170,000 employees, 140,000 
miles of track in operation, and over 1.3 million freight rail cars. 
Overall, FRA inspects about 0.2 percent of railroad operations each 
year. Its goal is to inspect all railroads at least once a year, but 
it does not always assess a railroad's compliance with all activities 
related to the requirements in each discipline during each inspection. 

When the major hours of service changes in RSIA took effect in July 
2009, the nation was in the midst of a serious economic recession, and 
the railroad industry was experiencing decreases in revenues, traffic, 
and staffing levels. For example, operating revenues for class I 
railroads decreased from $61.2 billion in 2008 to $47.8 billion in 
2009 before recovering to $58.4 billion in 2010.[Footnote 29] Revenue 
ton-miles[Footnote 30] for class I railroads followed a similar 
pattern, decreasing from 1.8 billion in 2008 to 1.5 billion in 2009 
before increasing to 1.7 billion in 2010. In addition, the number of 
class I railroad T&E employees decreased from about 65,000 in December 
2008 to just under 57,000 in December 2009 before increasing again to 
about 62,000 in December 2010. 

RSIA's Requirements Increased Rest Time and Decreased Fatigue Risk 
While Leaving Opportunities for Further Reductions in Fatigue Risk 
from Night Work: 

New Hours of Service Requirements Have Led to Changes in Covered T&E 
Employees' Work Schedules and Increased Rest Opportunity While 
Decreasing Hours Worked: 

RSIA's new hours of service requirements have led to changes in 
covered T&E employees' work schedules. Both the limits on consecutive 
work days without required rest (referred to hereafter as consecutive 
work day limits) and the new requirements for rest, including the 
requirements for increasing the minimum rest at the end of a shift 
from 8 to 10 hours, and for this rest to occur during the 24 hours 
before the start of a new shift and be undisturbed (referred to 
hereafter as the increased rest requirements) have contributed to the 
schedule changes. Factors other than RSIA, such as the economic 
recession, could have played a role in these changes. We attempted to 
mitigate for the effects of the economic conditions by avoiding 
choosing months during which the demand for rail service was rapidly 
declining and we also only analyzed employees that worked in both May 
2008 and May 2010 under the assumption that these employees would 
likely be performing similar work. 

Class I and II railroad officials we spoke with said RSIA's 
consecutive work day limits have led some railroads generally to 
substitute a schedule with 5 consecutive work days followed by 48 
hours of rest, known as a "5 by 2" schedule, for the previously more 
common schedule with 6 consecutive work days followed by 24 hours of 
rest, known as a "6 by 1" schedule. Now, use of the 6 by 1 schedule 
requires an FRA-approved waiver of compliance with hours of service 
requirements.[Footnote 31] For affected employees, this schedule 
change means that during the course of a 7-day period, a day of rest 
has taken the place of a day of work. RSIA's requirements for 
increased rest also contributed to the schedule changes. Although RSIA 
made 10 hours of rest mandatory, some railroad officials we spoke with 
said they had instituted 10-hour rest periods for covered T&E 
employees before RSIA took effect. However, this policy generally 
applied only at home terminals, not at away-from-home terminals. 
Railroad officials told us the work schedule changes responded to RSIA 
provisions and also addressed economic factors. 

Our analysis of hours worked by the same class I covered T&E employees 
and covered T&E employees of the participating class II railroads 
showed a per-employee increase of about 10 hours in the time available 
for rest for the class I employees in May 2010 compared with May 2008 
and a per-employee increase of about 17 hours for the class II 
employees in May 2010 compared with May 2008.[Footnote 32] This 
increase is statistically significant.[Footnote 33] The extent to 
which covered employees used the additional time to rest is unknown. 
Railroad officials told us that some employees have used the extra 
rest time to work a second job or to do other activities that may not 
involve rest. For example, an official with a class III railroad told 
us many of its covered T&E employees have farms that they work when 
they are not working on the railroad. 

The increased time available for rest under RSIA also led covered T&E 
employees to work fewer hours. The same analysis that we used to 
determine the increase in available rest time showed the same per- 
employee decrease in hours worked in May 2010 compared with May 2008-- 
about 10 hours for class I employees and about 17 hours for the 
selected class II employees--both of which are a statistically 
significant change. For both the class I and class II covered T&E 
employees included in our analysis, the total hours worked per 
employee decreased from 156 in May 2008 to about 146 in May 2010 for 
class I employees and from 169 in May 2008 to about 153 in May 2010 
for class II employees. In addition, for class I covered T&E employees 
the total number of work shifts (which includes covered and noncovered 
service for the railroad) per employee decreased from 18 in May 2008 
to 17 in May 2010 and class II covered T&E employees saw a decrease 
from 19 in May 2008 to 18 in May 2010. 

Still another effect of RSIA's increase in rest time may be an 
increase in the amount of time some covered T&E employees spent at 
terminals other than their home terminal. In responding to our rail 
industry survey, 6 out of 7 class I railroads reported an increase in 
the time affected employees spent at away-from-home terminals. In 
addition, 7 of 14 class II railroads reported that they had away-from-
home operations, and of these, 4 reported an increase in the time 
spent away from home. For the most part, increased time spent away 
from home was not an issue for the 153 class III railroads that 
responded to this question in our survey. Of these 153 class III 
railroads, 10 reported an increase in time at away-from-home terminals 
for their covered employees. According to both class I railroad and 
rail labor officials we spoke with, some of the affected covered 
employees are not happy with the increased time away-from-home, and 
the officials suggested that the undisturbed rest requirement be 
reduced from 10 hours to 8 hours at away-from-home terminals to allow 
covered employees to return home sooner. 

Hours of Service Changes Have Generally Reduced Fatigue Potential for 
Covered T&E Employees: 

Initial indications are that RSIA's changes generally reduced the 
fatigue potential for covered T&E employees. According to our analysis 
of covered T&E employee work schedules, the potential for covered 
employees to work at high risk of fatigue--a level associated with 
reduced alertness and an increased risk of errors and accidents-- 
decreased after RSIA took effect.[Footnote 34] More specifically, our 
analysis of the May 2008 and May 2010 work schedules for class I and 
class II covered T&E employees using an FRA-validated fatigue 
model[Footnote 35] showed that the percentage of total time worked at 
high risk of fatigue decreased by 29 percent (3 percentage points) for 
the class I employees and 36 percent (5 percentage points) for the 
class II employees (see figure 4).[Footnote 36] Further information on 
fatigue science and our use of fatigue models appears in appendix II. 

Figure 4: Class I and Class II Railroad High Fatigue Risk Level 
Changes for Covered T&E Employees Based on Total Hours Worked, May 
2008 and May 2010: 

[Refer to PDF for image: 2 horizontal bar graphs] 

Class I railroads: 
May 2008: 
Percent of time with high fatigue risk: 10.7%; 
May 2010: 
Percent of time with high fatigue risk: 7.6% (29% reduction). 

Class II railroads: 
May 2008: 
Percent of time with high fatigue risk: 12.6%; 
May 2010: 
Percent of time with high fatigue risk: 8.1% (36% reduction). 

Source: GAO analysis of rail workforce data. 

[End of figure] 

RSIA's consecutive work day limits and requirement for 10 hours 
undisturbed rest time both may have contributed to the reductions in 
work-related fatigue indicated by our analysis. 

* Effects of consecutive work day limits on fatigue. In its March 10, 
2011, Notice of Proposed Rulemaking Regulatory Impact Analysis on 
commuter and intercity passenger rail hours of service requirements, 
FRA stated that working an increasing number of consecutive days tends 
to result in reduced sleep as an employee sacrifices time for sleep to 
attend to personal activities. This tendency would apply to both 
freight and passenger railroads. FRA's proposed requirements for 
limiting consecutive days of work for commuter and intercity passenger 
rail covered T&E employees are based on research from other industries 
that shows some evidence of increased fatigue risk over successive 
workdays. In FRA's view, the proposed consecutive work day limits for 
commuter and intercity passenger rail covered employees were 
reasonable and necessary because of the increased fatigue risk from 
working a high number of consecutive days without rest.[Footnote 37] 
Rail labor representatives we spoke with also told us they see RSIA's 
consecutive work day limits as beneficial because they provide a break 
for employees in their work schedules. 

* Effects on fatigue of RSIA's increased rest requirements. As noted 
above, the total time available for rest has increased offering more 
opportunity for employees to rest. In addition RSIA's requirement that 
rest be undisturbed may be an additional benefit for employees rest. 
According to rail labor representatives we spoke with, crew calls from 
railroads during employees' rest periods were a concern before RSIA 
took effect. The representatives told us that covered employees had 
complained to them about unnecessary contact by railroads during their 
rest periods and said that this contact had been disruptive to their 
rest. Since RSIA has taken effect, they said, such complaints have 
virtually ceased. It is still early in the implementation process, yet 
class I railroad officials and rail labor officials we spoke with said 
that RSIA's requirements, including longer, undisturbed rest time 
should contribute to a better rested workforce. 

The effects of RSIA's hours of service changes on fatigue levels for 
class III railroads may depend on their operations. We did not analyze 
class III covered T&E employee work schedules, since their hours of 
service and employee records were largely paper-based. However, 
interviews with class III railroads indicated that for some class III 
railroads, particularly those that had scheduled daytime operations, 
fatigue may not have been an issue prior to RSIA. In interviews, 
railroad officials with 2 class III railroads said that fatigue was 
not an issue for their employees, because they offered service Monday 
through Friday during the daytime, with occasional Saturday service, 
depending on customer needs. Both of these railroads had FRA-approved 
waivers of compliance with hours of service requirements that 
permitted 6 by 1 work schedules, so that periodically scheduling a 
sixth day of service was not a concern. In addition, according to 
these officials, their covered T&E employees generally travel a 
maximum of 25 to 50 miles, and their work schedules always begin and 
end at the home terminal. In responding to our rail industry survey, 
98 out of 153 (64 percent) of the class III railroads responded that 
they had changed crew schedules as a direct result of RSIA. This 
change in crew schedules may indicate previous crew schedules were not 
compliant with RSIA provisions and changes to the schedules could have 
improved fatigue. Again, any improvements would likely be due in part 
to RSIA's consecutive work day limits and increased rest requirements. 

Opportunities Remain to Address Fatigue Risks from Night Work: 

Fatigue science has shown that the risk of fatigue is greater for 
nighttime work than for daytime work. For example, research on human 
circadian rhythms (the natural wake and sleep patterns of the human 
body) has shown that people by nature get tired at night and are more 
likely to have higher quality, more restorative sleep at night than 
they are during the day. Working at night can upset these circadian 
rhythms and result in sleep disruption and potential health problems. 
[Footnote 38] Fatigue research has also shown that fatigue increases, 
or alertness and performance decrease, during night work and that 
fatigue risk is substantially greater for successive night shifts than 
for successive day shifts.[Footnote 39] 

Eliminating nighttime work in the freight railroad industry would not 
be practicable, and RSIA's requirements had little effect on the 
amount of time covered T&E employees work at night. According to our 
analysis of the May 2008 and May 2010 work schedules for covered class 
I and class II T&E employees, the number of per-employee work hours 
that occurred at night for these employees decreased 4 hours per 
employee for class I employees and 2 hours for class II employees 
after RSIA took effect.[Footnote 40] In general, the freight railroad 
industry operates every day of the year, 24 hours a day. FRA noted in 
its Regulatory Impact Analysis for the proposed commuter and intercity 
passenger rail hours of service rules, that unlike freight service 
passenger service may be less affected by night work fatigue factors 
because most scheduled commuter and intercity passenger rail service 
does not operate during night hours. Additionally, our analysis of the 
work schedules for these 2 months showed little change in the 
percentage of covered class I T&E employees whose schedules involved 
night work--47 percent in May 2008 and 45 percent in May 2010. For the 
covered class II T&E employees, that percentage was the same in both 
months--34 percent. 

Although fatigue science has shown that the risk of fatigue is higher 
for night work than for day work, RSIA does not differentiate between 
the two in its hours of service requirements for freight railroads. 
FRA, however, has differentiated between the two when approving 
petitions for waivers of compliance with hours of service 
requirements. For example, when railroads have petitioned for a waiver 
of compliance with hours of service requirements to allow their 
employees to work a 6 by 1 schedule with both day and night shifts, 
FRA has approved such a schedule for daytime shifts, but not for 
shifts that include the hours between midnight and 6 a.m. FRA does not 
approve these shifts because of the higher risk of fatigue associated 
with them. FRA has also differentiated between day and night work in 
the final hours of service rules for covered train employees providing 
commuter and intercity passenger rail transportation. Specifically, 
the rule would not require FRA review and approval, including an 
assessment of fatigue risk, for work schedules that fall within the 
parameters of preapproved daytime work schedule templates (generally 
between 4 a.m. and 8 p.m.). Schedules that include work between 8 p.m. 
and 4 a.m., must generally be analyzed using an FRA approved fatigue 
model to assess the potential fatigue risk. FRA review and approval is 
needed on schedules where fatigue risk is deemed too great. For such 
schedules, railroads must generally take mitigating action to bring 
the risk from fatigue to an acceptable level. Additionally, 
limitations are placed on the number of consecutive days that a 
covered commuter or intercity passenger railroad T&E employee may 
work, with the limitations depending on the time of day of the 
assignments within the series of consecutive days. In making this 
distinction between nighttime and daytime work assignments, FRA has 
taken into account the fact that work at night presents a greater risk 
of fatigue.[Footnote 41] 

Our analysis of the class I and selected class II covered T&E employee 
work schedules for May 2008 and May 2010 shows that the extent that 
employees worked hours at night was highly correlated with employees 
spending at least 20 percent or more of their work time at high risk 
of fatigue.[Footnote 42] In our analysis the proportion of employees 
with 20 percent or more of their work time at high risk of fatigue for 
class I T&E covered employees decreased from 14 percent in May 2008 to 
10 percent in May 2010 and from 18 percent in May 2008 to 12 percent 
in May 2010 for participating class II railroad covered T&E employees. 
Even though fatigue risk was reduced after the implementation of the 
new hours of service requirements under RSIA, our findings on the 
correlation between night work and work hours spent at high risk of 
fatigue--along with the fatigue model results discussed previously 
that showed the decline in high risk of fatigue based on total hours 
worked--indicate that because RSIA did not directly limit the hours 
worked at night or incorporate night work into the freight 
requirements, fatigue might not be addressed under the new 
requirements to the fullest extent possible. Taking hours worked at 
night into consideration for freight hours of service requirements 
could hold promise for mitigating the risk of fatigue. 

In addition to analyzing actual work schedules from the class I and 
class II railroad T&E employees, we analyzed three consecutive sets of 
two hypothetical 6 by 2 work schedules--the maximum number of 
consecutive work days allowed under RSIA when not returning home from 
an away-from-home terminal--using a fatigue model to further assess 
the effects of night work on fatigue. One schedule included only 
daytime hours with 10 hour shifts from 8 a.m. to 6 p.m. and the other 
included nighttime hours with 10 hour shifts from 8 p.m. to 6 a.m. 
According to our analysis, the percentage of time at high risk of 
fatigue was greater for the hypothetical night work schedule than for 
the hypothetical day work schedule. The day work schedule had no time 
spent at high risk of fatigue, while the night work schedule had a 
total of 67 hours, or 37 percent of total work time, spent at high 
risk of fatigue. Furthermore, the risk of fatigue for the nighttime 
work schedule was high for all but one of the work days in the all-
night-work schedule, while no work day in the all-day-work schedule 
fell into the high risk category (see figure 5). The peak fatigue 
score shown in the figure is the highest fatigue score achieved on a 
work schedule day analyzed by the fatigue model. This does not mean 
the whole scheduled work time was spent working at the peak fatigue 
level. For example, on the nighttime hours schedule on day 2 according 
to the model example employee 2 would have spent 32 minutes working at 
high risk of fatigue with a peak fatigue score of 72. However, 
according to the model output on day 6 example employee 2 spent 8 
hours 6 minutes out of a 10 hour shift working at high risk of fatigue 
with a peak fatigue score of 119. As our analysis of these 
hypothetical schedules indicates, consecutive daytime shifts may 
present a lower risk of fatigue than consecutive nighttime shifts. 

Figure 5: Differences in Fatigue Risk for Hypothetical Daytime and 
Nighttime Work Schedule: 

[Refer to PDF for image: multiple line graph] 

FAID score: 

High risk of fatigue threshold (70). 

Day 1: 
Employee 1 peak FAID score per day (working day shifts): 34; 
Employee 2 peak FAID score per day (working night shifts): 56. 

Day 2; 
Employee 1 peak FAID score per day (working day shifts): 40; 
Employee 2 peak FAID score per day (working night shifts): 72. 

Day 3: 
Employee 1 peak FAID score per day (working day shifts): 46; 
Employee 2 peak FAID score per day (working night shifts): 86. 

Day 4: 
Employee 1 peak FAID score per day (working day shifts): 52; 
Employee 2 peak FAID score per day (working night shifts): 99. 

Day 5: 
Employee 1 peak FAID score per day (working day shifts): 58; 
Employee 2 peak FAID score per day (working night shifts): 111. 

Day 6: 
Employee 1 peak FAID score per day (working day shifts): 64; 
Employee 2 peak FAID score per day (working night shifts): 119. 

Day 7: 
Employee 1 peak FAID score per day (working day shifts): 0; 
Employee 2 peak FAID score per day (working night shifts): 0. 

Day 8: 
Employee 1 peak FAID score per day (working day shifts): 0; 
Employee 2 peak FAID score per day (working night shifts): 0. 

Day 9: 
Employee 1 peak FAID score per day (working day shifts): 34; 
Employee 2 peak FAID score per day (working night shifts): 74. 

Day 10: 
Employee 1 peak FAID score per day (working day shifts): 40; 
Employee 2 peak FAID score per day (working night shifts): 83. 

Day 11: 
Employee 1 peak FAID score per day (working day shifts): 46; 
Employee 2 peak FAID score per day (working night shifts): 93. 

Day 12: 
Employee 1 peak FAID score per day (working day shifts): 52; 
Employee 2 peak FAID score per day (working night shifts): 102. 

Day 13: 
Employee 1 peak FAID score per day (working day shifts): 58; 
Employee 2 peak FAID score per day (working night shifts): 111. 

Day 14: 
Employee 1 peak FAID score per day (working day shifts): 64; 
Employee 2 peak FAID score per day (working night shifts): 119. 

Day 15: 
Employee 1 peak FAID score per day (working day shifts): 0; 
Employee 2 peak FAID score per day (working night shifts): 0. 

Day 16: 
Employee 1 peak FAID score per day (working day shifts): 0; 
Employee 2 peak FAID score per day (working night shifts): 0. 

Day 17: 
Employee 1 peak FAID score per day (working day shifts): 34; 
Employee 2 peak FAID score per day (working night shifts): 74. 

Day 18: 
Employee 1 peak FAID score per day (working day shifts): 40; 
Employee 2 peak FAID score per day (working night shifts): 83. 

Day 19: 
Employee 1 peak FAID score per day (working day shifts): 46; 
Employee 2 peak FAID score per day (working night shifts): 93. 

Day 20: 
Employee 1 peak FAID score per day (working day shifts): 52; 
Employee 2 peak FAID score per day (working night shifts): 102. 

Day 21: 
Employee 1 peak FAID score per day (working day shifts): 58; 
Employee 2 peak FAID score per day (working night shifts): 111. 

Day 22: 
Employee 1 peak FAID score per day (working day shifts): 64; 
Employee 2 peak FAID score per day (working night shifts): 119. 

Day 23: 
Employee 1 peak FAID score per day (working day shifts): 0; 
Employee 2 peak FAID score per day (working night shifts): 0. 

Day 24: 
Employee 1 peak FAID score per day (working day shifts): 0; 
Employee 2 peak FAID score per day (working night shifts): 0. 

Source: GAO analysis of example schedule data. 

Note: We used the FAID fatigue model to analyze fatigue risk for these 
hypothetical work schedules. 

[End of figure] 

New Hours of Service Requirements Have Led to Operational and 
Administrative Changes That Have Increased Some Railroads' Costs: 

Operational Changes Have Affected Railroads' Scheduling, Staffing, and 
Customer Service: 

According to our survey results, RSIA's hours of service requirements 
led to a number of effects on railroads' operations, as would be 
expected with any significant change in statutory or regulatory 
requirements aimed at improving safety by reducing covered employee 
fatigue. These effects included changes in how crews and trains are 
scheduled, increases in staffing levels to maintain operations, and 
reductions in some railroads' ability to meet customer needs. In 
general, according to our survey results, smaller railroads found some 
of the changes more burdensome than did larger railroads. In addition, 
some railroads incurred one-time or ongoing financial costs, or both, 
to implement the changes. 

Hours of Service Changes Have Led to Alterations in Railroads' Crew 
and Train Scheduling and Increased Some Railroads' Costs: 

According to our survey results, RSIA's hours of service requirements-
-especially its consecutive work day limits and increased rest 
requirements--substantially changed the way railroads schedule crews. 
For example, all 7 class I, 8 of 14 class II (about 57 percent), and 
98 of 152 class III (about 64 percent) railroads reported changing 
crew schedules as a direct result of RSIA's hours of service 
requirements.[Footnote 43] Such changes would be expected, given the 
new requirements. Prior to RSIA, covered T&E employees on some 
railroads often worked well beyond 6 or 7 consecutive days. Officials 
we spoke with at 1 class I railroad said its train crews often worked 
8 consecutive days followed by 3 days off, and officials at another 
class I railroad said most of its employees worked 6 consecutive days 
with 1 day off, although covered employees often worked 7 days 
followed by 3 days off or 11 days followed by 4 days off. According to 
officials we interviewed at 1 class II railroad, a small portion of 
its covered train employees (about 15 percent) worked up to 22 
consecutive days followed by 8 days off. After RSIA, covered employees 
could no longer work for more than 6 or 7 consecutive days without 
taking required rest. According to railroad officials, the requirement 
for 48 hours' rest following 6 consecutive work days has been 
particularly challenging, and some officials told us they try to avoid 
working employees 6 consecutive days. 

Our survey results indicate that the changes in crew schedules led to 
changes in train schedules. Specifically, in responding to our survey, 
4 of 7 class I railroads reported changing train schedules as a direct 
result of RSIA's hours of service changes, while 5 of 14 (about 36 
percent) class II railroads and just under half (70 of 153) of class 
III railroads reported making this change. Changes in train schedules 
particularly affected smaller railroads. According to officials we 
interviewed from 1 class II railroad, RSIA's changes, particularly the 
additional time needed for employees to return to work, made it 
difficult to maintain train schedules and to respond to changes in 
train operating plans, which are often caused by factors such as 
mechanical problems and traffic levels. RSIA's changes reduced their 
flexibility in such situations. An official from another class II 
railroad told us that RSIA's hours of service changes meant the 
railroad had to reduce train service from 7 days a week to 6 days 
because it did not have enough people available to offer service 7 
days a week. While this railroad has since hired people and said it 
expects to resume 7-day service, it was not able to do so for over a 
year. In some instances, train connections were also affected. For 
example, a class II railroad official we interviewed said that delays 
on some of the company's long-distance trains, which the official 
attributed to RSIA's changes, led to delays on local trains that 
connected with the long-distance trains. Officials from another class 
II railroad said RSIA's changes caused them to hold trains out of 
their rail yard because, until March 2011, they did not have enough 
people to handle them. 

While RSIA's consecutive day limits and increased rest requirements 
were focused on reducing fatigue and improving safety, a majority of 
the railroads responding to our survey reported that the resulting 
changes in crew and train schedules imposed burdens on them, and some 
of these railroads reported that the changes increased their costs. As 
shown in figure 6, the burden on railroads from changing train 
schedules could be very great, especially for smaller railroads. Three 
of the 4 class I railroads responding to this survey question reported 
a moderate to substantial burden, and over 55 percent of the 
responding class II and III railroads reported a substantial to very 
great burden. For example, as officials we interviewed from a holding 
company that owns over 30 smaller railroads said, the burden imposed 
by changing crew and train schedules was very great for some 
railroads--such as those that serve grain producers during the harvest 
season--that need to run trains without interruption at certain 
seasons to meet demand. According to the officials, RSIA's consecutive 
work day limits and requirement for more rest between shifts make 
uninterrupted service like this very difficult to provide. Finally, 
according to our survey results, changes to crew and train schedules 
entailed financial costs, particularly for class I railroads. Of the 7 
class I railroads, 5 reported incurring financial costs from changing 
crew schedules and 4 reported incurring such costs from changing train 
schedules. Fewer class II and III railroads reported incurring such 
costs, although half (76 of 152) of the class III railroads that 
reported changing crew schedules reported incurring financial costs 
for doing so. In some cases, these may have been one-time costs, such 
as for upgrading hours of service timekeeping systems to accommodate 
new crew schedules. In other cases, they may have been recurring 
costs, such as for hiring new employees or bringing employees back 
from furlough to address issues related to crew or train schedules 
(discussed later in this report). Costs for additional staff could 
also be related to service increases responding, at least in part, to 
improvements in the economy that followed RSIA's implementation in 
July 2009. 

Figure 6: Degree of Burden on Surveyed Railroads from Changing Train 
Schedules As a Direct Result of RSIA's Hours of Service Changes: 

[Refer to PDF for image: horizontal bar graph] 

How much of an operational burden - Change train schedules: 

Number of railroads responding: 

Class I: 
Little or no burden: 0; 
Some burden: 1; 
Moderate burden: 2; 
Substantial burden: 1; 
Very great burden: 0. 

Class II: 
Little or no burden: 0; 
Some burden: 1; 
Moderate burden: 1; 
Substantial burden: 2; 
Very great burden: 1. 

Class III: 
Little or no burden: 2; 
Some burden: 11; 
Moderate burden: 18; 
Substantial burden: 23; 
Very great burden: 16. 

Source: GAO survey. 

[End of figure] 

In general, we did not ask the railroads we surveyed to identify 
specific dollar amounts incurred as a result of RSIA's hours of 
service changes or to indicate how those amounts may have affected 
railroad earnings or profits. We did ask the railroads to identify the 
average annual wages and benefits for employees hired or brought back 
from furlough as a result of RSIA's changes (discussed later in this 
report). During our interviews, some railroad officials told us it was 
difficult to separate the financial effects of RSIA's changes from 
those of general economic conditions. Nevertheless, even though we did 
not determine the specific financial effects of RSIA's changes on 
railroads, it is likely the changes affected the costs, revenues, and 
earnings for some railroads, at least temporarily. As discussed 
earlier, such effects are not unexpected given the magnitude of RSIA's 
hours of service changes and the many actions required by railroads to 
comply with the law. 

Staffing Levels for Some Railroads Increased and There Were Other 
Effects on Staffing As Well: 

In implementing RSIA's hours of service changes to improve safety and 
comply with the law, some railroads reported increasing their staffing 
levels in response to the changes they made in crew schedules. In 
general, according to railroad officials we spoke with, staffing 
levels increased with the changes in crew schedules because, with 
RSIA's consecutive work day limits and increased rest requirements, 
covered T&E employees were less available to work. More specifically, 
because of RSIA's requirements for 48 instead of 24 hours' rest after 
6 consecutive days on duty, and for 10 instead of 8 hours' rest 
between shifts, employees were generally less available for work and 
more staff were needed to maintain regular operations. For example, 
officials we spoke with from a class II railroad said RSIA's 
requirements for 10 hours' rest between shifts, and for this rest to 
be undisturbed, could increase the time that covered employees were 
unavailable for work by between 2 and 4 hours and meant, for this 
railroad, that more staff were needed to provide pre-RSIA service 
levels.[Footnote 44] Additionally, according to an official we spoke 
with from a class I railroad, RSIA's changes meant that this railroad 
needed about 200 more T&E employees than it previously did to run the 
same amount of business. Although we tried to isolate RSIA's effects 
on railroads' staffing by asking railroads to identify the extent to 
which they hired new employees or brought employees back from furlough 
as a direct result of RSIA, we cannot exclude the possibility that 
some of the changes they reported were also due to improvements in 
general economic conditions that took place from 2009 to 2010. 

To address staffing needs, railroad officials we spoke with told us 
they called on T&E employees without regular crew assignments, hired 
new employees, or brought employees back from furlough to help fill 
the staffing gaps. T&E employees without regular crew assignments are 
listed on what are called extra boards and are on call to meet crew 
needs as they arise, giving railroads flexibility to meet staffing 
needs when regular crews are not available to work. All 7 class I 
railroads use extra boards, and some smaller railroads may also use 
them. In addition, railroads reported hiring new people or bringing 
people back from furlough. Some railroad officials we spoke with said 
these people were at least initially assigned to extra boards. In 
responding to our survey, 5 of 7 class I and 7 of 14 class II 
railroads reported they hired or brought T&E employees back from 
furlough as a direct result of RSIA's requirements. Proportionally, 
fewer small railroads reported hiring or bringing employees back from 
furlough--about 30 percent (46 of 152) of the class III railroads 
responding to our survey. In some instances, hiring decisions at 
smaller railroads may have reflected broader economic conditions 
rather than specific operating needs. For example, an official we 
spoke with from a class III railroad said the company was apprehensive 
about long-term hiring because, given the risks of a sudden decline in 
orders, it might have to lay employees off after investing in their 
hiring and training. 

According to our survey results, the number of T&E employees railroads 
hired or brought back from furlough varied and increased some 
railroads' costs. Overall, as would be expected, larger railroads 
reported hiring or bringing back more employees than smaller 
railroads. For those railroads we surveyed that reported hiring or 
bringing people back from furlough, the number of people ranged 
between 120 and 500 each for the 4 class I railroads, 5 and 40 each 
for the 7 class II railroads, and 1 and 30 each for the 45 class III 
railroads. In hiring or bringing T&E employees back from furlough, the 
railroads incurred ongoing financial costs. According to our 
estimates, based on the average annual wages and benefits of T&E 
employees reported by the railroads we surveyed, the average annual 
cost for the 4 class I railroads ranged from about $11 million to $50 
million, and for the 7 class II railroads, it ranged from about 
$350,000 to $3 million. 

While RSIA's requirements affected some railroads' need for staff, the 
requirements had other effects on staffing as well, including reduced 
flexibility in using managers and reduced ability to provide 
guaranteed and other work hours to covered employees: 

* Reduced flexibility to use managers to perform covered and 
noncovered service.[Footnote 45] Our survey results indicated that 
RSIA's changes may have reduced the ability of some managers to 
perform covered and noncovered service. Most of the larger railroads 
we surveyed--5 of 7 class I railroads and 12 of 14 class II railroads--
reported no reductions in the ability of managers to perform covered 
service. In contrast, about 36 percent (54 of 151) of class III 
railroads reported a reduction in managers' ability to perform covered 
service, and about 30 percent (46 of 151) reported a reduction in 
managers' ability to perform noncovered service. In general, this 
issue is of particular importance for smaller railroads. ASLRRA 
officials told us, that on small railroads, the same person often 
performs many different functions and it is not unusual for railroad 
managers to operate trains in place of employees who are sick or on 
vacation in addition to performing their managerial responsibilities. 
However, the officials said, if managers do perform such work, they 
come under RSIA's hours of service limitations, including the monthly 
cap on total work hours. All of their work hours, including the time 
spent performing both covered and noncovered service, then falls under 
the 276-hour cap. The officials said, in some instances, this 
restriction could prevent managers from performing their regular 
managerial work. 

* Reduced ability to provide guaranteed work hours to covered 
employees. Some railroads, generally by collective bargaining 
agreement, guarantee a minimum number of work hours or days to 
employees over a certain period (e.g., 2 weeks). In some cases, the 
railroads do this to retain a certain class of employee, such as T&E 
employees. In general, employees are paid for the guaranteed hours or 
days whether they perform the work or not. For a railroad, not 
providing work during the guaranteed hours may mean having to pay for 
work not performed. For an employee, not meeting guaranteed hours or 
days may mean fewer hours worked, even though the employee may be paid 
for the time not worked. In responding to our survey, 4 of the 7 class 
I railroads reported they were not able to meet guaranteed work hours 
as a direct result of RSIA's requirements, whereas smaller proportions 
of class II and III railroads reported this issue (see figure 7). 

Figure 7: Number of Surveyed Railroads That Did Not Meet Guaranteed 
Hours As a Direct Result of RSIA's Requirements, by Class of Railroad: 

[Refer to PDF for image: horizontal bar graph] 

Action occurs as direct result of the RSIA hours of service 
requirements - Guaranteed hours not met: 

Number of railroads responding: 

Class I: 
Yes: 4; 
No: 3. 

Class II: 
Yes: 5; 
No: 9. 

Class III: 
Yes: 22; 
No: 131. 

Source: GAO survey. 

[End of figure] 

* Potentially reduced ability to provide work hours to covered 
employees each month. RSIA's monthly cap on total work hours (276 
hours) may have altered how railroads are able to use their workforce 
and the number of hours employees work. A number of railroad officials 
we interviewed said this cap did not affect their company. However, 
the majority of railroads we surveyed reported taking employees 
temporarily out of service as a direct result of RSIA's hours of 
service changes.[Footnote 46] This may have resulted from controls 
railroads implemented to prevent covered T&E employees from exceeding 
the total monthly work hour cap. For example, in follow-up work on our 
survey, we learned that all 7 class I railroads established internal 
thresholds to monitor employee work hours to ensure that employees did 
not exceed this cap. These thresholds ranged from 250 to 264 hours. 
[Footnote 47] We did not determine through our survey and interviews 
how many T&E employees may have been taken out of service because of 
these thresholds. However, the number of employees reported by our 
survey respondents as reaching or exceeding the total monthly work 
hour cap in any particular month in 2010 was small and ranged from 0 
to 26 for all railroads surveyed.[Footnote 48] Whether or to what 
extent the internal thresholds influenced this number, prevented 
employees from exceeding the monthly cap, or limited work hours is 
unknown. 

Some of the labor organizations we spoke with expressed concerns about 
how railroads responded to RSIA's hours of service requirements. For 
example, officials we interviewed from one organization that 
represents T&E employees said that a significant portion of its 
membership had suffered some salary loss from the changes in hours of 
service requirements and added that covered employees were not being 
scheduled for more than 252 hours of work in a month in order to avoid 
reaching or exceeding RSIA's cap on total monthly work hours. The 
officials said this practice can cost employees as much as 24 hours' 
pay in a month. Officials from unions representing conductors, 
signalmen, and yardmasters[Footnote 49] expressed similar concerns. 
They primarily attributed reductions in work hours and lost 
compensation to RSIA's impact on crew schedules as well as to the 
requirement for 10 hours' undisturbed rest and the monthly work hour 
cap. For example, an official with a union representing yardmasters 
told us the requirement for 10 hours' undisturbed rest precludes 
employees that work in rail yards from working swing (third) shifts as 
well as regular shifts 7 days a week and this restriction deprives 
employees of work and reduces earning opportunities. Union officials 
also told us the internal threshold some railroads use to address the 
monthly work hour cap serves as an artificial cap and essentially 
deprives employees of additional work hours and earnings. 

Hours of Service Changes May Have Affected Some Railroads' Ability to 
Meet Customer Needs: 

The changes that some railroads made to implement RSIA's hours of 
service changes and improve safety may also have limited their ability 
to provide service and meet customer needs. As figure 8 shows, over 
half of all railroads (98 of 174 railroads) responding to our survey 
question reported their ability to meet customer needs was reduced as 
a direct result of RSIA's hours of service changes. In particular, 
class I (4 of 7) and class III (87 of 153) railroads reported a 
reduction. Railroad officials we spoke with largely attributed these 
effects to RSIA's consecutive work day limits and requirements for 
increased rest. By affecting crew and train schedules, the officials 
noted, the requirements have sometimes acted to limit railroads' 
flexibility to provide train service when and where needed, especially 
on weekends. 

Figure 8: Railroad Industry Responses about Effects on Customer 
Service As a Direct Result of RSIA Hours of Service Changes, by Class 
of Railroad: 

[Refer to PDF for image: horizontal bar graph] 

Action occurs as direct result of the RSIA hours of service 
requirements - Ability to meet customer needs was reduced: 

Number of railroads responding: 

Class I: 
Yes: 4; 
No: 3. 

Class II: 
Yes: 7; 
No: 7. 

Class III: 
Yes: 87; 
No: 66. 

Source: GAO survey. 

[End of figure] 

We did not determine the effects of RSIA's requirements on changes in 
railroad customer service, such as whether railroads lost customers or 
customers changed modes of transportation following service changes. 
RSIA's focus was on improving railroad safety but effects on customer 
service may have occurred. Officials we spoke with at a class II 
railroad said that, in some instances, the hours of service 
requirements have led to about a 50 percent loss in weekend crew 
starts and negative effects on customer service. On weekends, the 
officials said, some entire industries do not receive train service 
because people are not available to operate the trains. Officials we 
spoke with at several other smaller railroads also told us their 
weekend service had been affected. For some railroads, reduced 
customer service may have been temporary. For example, some railroads 
have petitioned FRA for waivers of compliance from hours of service 
requirements so they will have the flexibility to provide service for 
6 days (e.g., Monday through Saturday), followed by 24 hours' rest 
rather than the RSIA-mandated 48 hours, if customer needs dictate. In 
addition, hiring new employees or bringing employees back from 
furlough may have permitted some railroads to return to service levels 
that may have decreased initially because the railroads lacked 
available employees. Officials at one class II railroad told us they 
believed their ability to keep up with business levels was severely 
hampered by the lack of covered employees caused by RSIA's hours of 
service changes. To address the issue, they brought back all their 
previously furloughed employees and hired even more covered employees. 
As a result, by March 2011, the officials believed the railroad was 
getting back to crew levels that were sufficient to meet business 
needs. 

Some shippers and receivers that use rail to meet their transportation 
needs told us their service had been affected by RSIA. We did not 
formally survey shippers or receivers that use rail to transport their 
goods about the possible effects of RSIA's hours of service changes, 
but responses to questions sent out on our behalf by a trade 
association (the National Industrial Transportation League) that 
represents shippers and receivers of a wide mix of commodities, 
including steel, paper, and agricultural products, indicated that the 
changes had affected some of them.[Footnote 50] Of the 28 shippers and 
receivers that responded to the questions, 10 said their service had 
been affected by RSIA's hours of service changes, and 7 said their 
weekend service had been affected. Among the problems with service 
cited by these shippers and receivers were less predictable service, 
train crew shortages, and switches missed because crews were 
unavailable or had "timed out on the clock."[Footnote 51] The 
responses to these customer service problems varied but included 
increasing rail fleets, switching to trucks to compensate for rail 
inefficiencies, and increasing inventory or adjusting or shutting down 
production schedules. 

Administrative Changes Involved Recordkeeping Systems and Claims for 
Lost Work Opportunities: 

To implement RSIA's hours of service changes, railroads also reported 
making administrative changes. For example, railroads reported 
modifying or creating new recordkeeping systems to account for time 
covered by hours of service requirements, spending more time reviewing 
hours of service records, and handling more claims for lost work 
opportunities. 

Changes Imposed Recordkeeping Burdens for Most Railroads and Increased 
Some Railroads' Costs: 

To accommodate RSIA's new hours of service requirements and create the 
records necessary to comply with the law, railroads of all sizes 
reported modifying their timekeeping systems or, in some cases, 
creating new systems. According to our survey results, most large 
railroads (all 7 class I railroads and 12 of 14 class II railroads) 
primarily reprogrammed or updated their existing timekeeping systems, 
which are generally electronic (see figure 9). Among other things, 
officials we spoke with at some railroads said they established ways 
to track employees' total work and limbo or deadhead hours in a month 
and, in some cases, incorporated alerts to prevent covered employees 
from being contacted during undisturbed rest periods. Designing a way 
to prevent contact was sometimes more difficult than expected because, 
as officials we spoke with at one class I railroad said, people other 
than crew schedulers, such as company doctors, security personnel, and 
payroll personnel, may try to contact a covered employee during a day, 
and the system has to preclude all such contacts during an undisturbed 
rest period. In contrast, many small railroads we surveyed reported 
creating new timekeeping systems. Over half (94 of 151) of the class 
III railroads reported creating new timekeeping systems. Officials we 
interviewed at some of the class III railroads said their companies 
have paper-based hours of service timekeeping systems but use 
electronic spreadsheets to track covered employees' hours of service. 
In some cases, the electronic spreadsheets were updated to keep track 
of such things as total monthly work hours. Additionally, according to 
some railroad officials we spoke with, their changes were sometimes 
part of a broader effort to better manage both hours of service and 
other aspects of their business, such as financial management. 

Figure 9: Changes Railroads, or Their Parent Companies, Made Related 
to Timekeeping Systems As a Direct Result of RSIA's Hours of Service 
Changes, by Class of Railroad: 

[Refer to PDF for image: horizontal bar graph] 

Change as direct result of the RSIA hours of service requirements -
Created a new hours of service time-keeping system/reprogrammed or 
updated an existing hours of service time-keeping system: 

Number of railroads responding: 

Class I, created: 
Yes: 3; 
No: 4; 
No response: 0; 
Don't know: 0. 

Class I, reprogrammed or updated: 
Yes: 7; 
No: 8; 
No response: 0; 
Don't know: 0. 

Class II, created: 
Yes: 5; 
No: 9; 
No response: 0; 
Don't know: 0. 

Class II, reprogrammed or updated: 
Yes: 12; 
No: 2; 
No response: 0; 
Don't know: 0. 

Class III, created: 
Yes: 94; 
No: 51; 
No response: 5; 
Don't know: 1. 

Class III, reprogrammed or updated: 
Yes: 69; 
No: 71; 
No response: 6; 
Don't know: 4. 

Source: GAO survey. 

[End of figure] 

To demonstrate compliance with RSIA's hours of service changes, some 
railroads reported spending more time preparing or reviewing hours of 
service records--work that officials said sometimes limited their 
ability to perform other tasks, such as operating their business. 
Survey respondents who addressed this question, including those from 
all 7 class I railroads and over 70 percent of class II and III 
railroads (10 of 14 and 115 of 152, respectively), reported the time 
required for recordkeeping or recordkeeping review increased as a 
direct result of RSIA's hours of service requirements. Not 
unexpectedly, the increased time to prepare or review hours of service 
records imposed burdens on railroads. In responding to our survey, 6 
of the 7 class I railroads reported the additional time for 
recordkeeping or recordkeeping reviews presented some to a moderate 
burden, while half (5 of 10) of the class II railroads and about 40 
percent (44 of 111) of the class III railroads responding to this 
question reported a substantial to a very great burden. Over time, the 
increased efforts to prepare and review hours of service records will 
likely become part of the normal routine of a railroad. In addition, 
creating such records is part of helping ensure compliance with the 
law and achieving its intended safety benefits. However, at least 
temporarily, some railroads we spoke with said the increased record 
preparation and review time affected how their business is operated. 
For example, officials we spoke with at 3 class III railroads, all 
with paper-based hours of service records, said the additional 
information that must be tracked for hours of service records was a 
burden and the time spent on tracking left less time for other 
activities, including running the railroad. This was one of these 
railroads' main issues with RSIA's changes. An official we spoke with 
at a class I railroad also told us RSIA's changes added an extra layer 
of reporting to the company's hours of service process, primarily to 
accommodate RSIA's total monthly work hour caps. 

Finally, in responding to our survey, some railroads reported that the 
timekeeping changes imposed financial costs. In some cases, these may 
have been one-time costs, and in others, they may have been recurring 
costs. According to our survey results, all 7 class I railroads, 12 of 
the 14 class II railroads, and 88 of 150 class III railroads incurred 
financial costs from introducing or revising hours of service records 
or recordkeeping systems. We did not collect information on the 
specific costs incurred. However, some railroad officials we 
interviewed said the costs ranged into the millions of dollars. 
According to officials from one class I railroad we spoke with, it 
spent about $3 million in 2009 for programming changes, including 
changes to its crew monitoring system. Officials we spoke with at 
another class I railroad told us it spent about $2 million for 
programming and upgrades, including converting from paper to 
electronic records for its signal employees. According to the 
officials, the cost was primarily for company employees, not a 
consultant, to do the reprogramming and was a one-time cost. At some 
other railroads, the costs were for work performed by a mix of in-
house staff and outside consultants. Some of the costs were recurring. 
For example, an official from a class III railroad told us his company 
spends an extra $500 a month for a manager to review and verify the 
accuracy of hours of service records. 

Railroads Also Reported Increases in Claims Filed for Lost Work Hours 
or Compensation: 

After RSIA took effect, some covered employees filed claims for lost 
work or compensation--that is, requests for payment for work hours or 
compensation lost because of RSIA's consecutive work day limits or 
other requirements. Such claims might arise when, for example, an 
employee who formerly worked a 6 by 1 shift could no longer do so 
because RSIA requires 48 hours' rest after 6 consecutive days on duty. 
In responding to our survey, 5 of 7 class I, 6 of 14 class II, and 22 
of 152 class III railroads reported that the number of claims for 
missed work opportunities (hours) or compensation increased as a 
direct result of RSIA's hours of service changes. The remaining class 
I and II railroads and 128 of the class III railroads (about 84 
percent) said either the number of such claims stayed the same or the 
issue was not applicable to them.[Footnote 52] We did not collect data 
on the number of claims filed. However, some railroad officials we 
spoke with said the number of claims filed doubled or tripled from the 
normal level. For example, a class II railroad official we spoke with 
estimated that the number of claims filed at his company each month 
for lost work hours increased from about 5 before RSIA took effect to 
10 to 15 afterwards. This official also said the number of claims 
subsequently went back to 2-3 per month. A class I railroad official 
told us T&E employees had filed over 500 claims at this company 
between July 2009 and May 2010, most of which the company was holding 
in abeyance until it had decided how to handle them. 

We do not know how many claims may have resulted in payments to 
employees or other forms of relief. As noted, one class I railroad we 
spoke with had not decided at the time of our review how to resolve 
the 500 claims filed by its employees, in part because the railroad 
was still considering the status of collective bargaining agreements 
in relation to RSIA's legal requirements. Officials from this railroad 
estimated each claim filed averaged approximately $200 and the 
railroad's potential liability in paying these claims was about 
$100,000. At other railroads, paying compensation may have been more 
routine. For example, an official we spoke with at a class II 
railroad, which was trying to avoid working employees 6 consecutive 
days, said the railroad had, in virtually every instance, paid claims 
for compensation filed by T&E employees who had been skipped over for 
work assignments because they were approaching 6 consecutive days of 
work. 

Too Soon to Assess FRA's Hours of Service Oversight, but 
Implementation of Pilot Project and Waiver Provisions Could Yield More 
Data to Address Fatigue: 

FRA Uses a Risk-Based Approach to Oversee Compliance with Hours of 
Service Requirements: 

To plan its oversight of railroads' compliance with hours of service 
requirements, FRA applies the same risk-based approach that it uses to 
assess compliance generally. This approach relies on a risk-based 
model that FRA implemented in 2006. The model analyzes FRA's 
inspection data, together with accident and incident data reported by 
the railroads through the Accident and Incident Reporting System, and 
then generates the National Inspection Plan (NIP), which is designed 
to target FRA's inspections at the greatest safety risks. The NIP 
allocates inspection resources for each FRA region by inspector 
discipline (such as operating practices and track),[Footnote 53] and 
FRA regions then assign resources to activities (such as hours of 
service and drug and alcohol control), within each discipline with 
input from inspectors familiar with each railroad's operations. In 
addition, FRA regional officials can modify the NIP's allocation of 
resources among disciplines based on local input, both initially and 
after 6 months. According to FRA headquarters and regional officials, 
decisions about how to allocate resources among inspection disciplines 
and activities are based on factors such as complaints, an inspector's 
knowledge of a railroad's operation at a given location, and the time 
and resources available to conduct inspections. This reliance on local 
input reflects FRA's views that regional officials and inspectors have 
detailed knowledge about railroads' operations that may not be 
captured in the data used to develop the NIP and that their input 
results in a stronger inspection plan than one based solely on data 
analysis. 

FRA incorporates data from its Accident and Incident Reporting System 
into its risk assessment model to help determine the relationship 
between noncompliance with safety requirements and risk. Specifically, 
FRA establishes codes for a wide range of violations or conditions, 
and when railroads report an accident or incident, they enter two 
codes into the system--one for the primary cause and the other for a 
contributing cause of the accident or incident. Neither hours of 
service violations nor, more broadly, fatigue are among the coded 
options that railroads can choose to enter. Instead, the options 
include a large number of actions or conditions that FRA considers 
potentially related to fatigue, such as "failure to release hand brake 
on cars" and "failure to comply with restricted speed." When FRA 
investigates an accident or incident with these codes entered as 
causes, it then attempts to determine whether fatigue was a factor. 
According to FRA, it does not have a code for hours of service 
violations because, in its experience, there is not necessarily a 
relationship between hours of service and fatigue--a fatigued 
individual can be in compliance with hours of service requirements or, 
conversely, a violation of hours of service requirements can occur 
without an individual being fatigued. According to FRA, it does not 
have a code for fatigue because it already collects information on 
fatigue when it investigates an accident or incident.[Footnote 54] 

It is Too Soon to Determine if FRA's Hours of Service Oversight Has 
Changed under RSIA or Needs to Change: 

It is too soon since RSIA was implemented to determine if the priority 
that FRA assigns to overseeing railroads' compliance with hours of 
service requirements has changed or should change. The new hours of 
service requirements did not take effect for freight railroads until 
July 16, 2009, and we collected the inspection data through September 
30, 2010, a span of 14 months. Hence, the period covered by our audit 
work is too short for us to identify any trends in inspection results 
or enforcement actions taken since RSIA's changes went into effect. 
Furthermore without trend information, there is little basis at the 
current time to know whether the priority FRA assigns to overseeing 
hours of service is best aligned with potential safety risks. 

FRA inspectors conduct hours of service inspections and complaint 
investigations to determine if covered employees have worked longer 
than limited by law. FRA inspectors also review railroads' hours of 
service recordkeeping to assess their compliance with FRA regulations 
that specify, for example, how and when the hours worked by covered 
employees are to be recorded. According to our analysis of FRA 
inspection data, FRA inspectors conducted somewhat fewer hours of 
service and hours of service recordkeeping inspections of freight 
railroads in fiscal year 2010--the one complete year for which we have 
data since RSIA took effect--than they did in fiscal year 2008, the 
last full year before RSIA was implemented. (See table 2.) However, 
the data for fiscal year 2010 show increases, especially for hours of 
service inspections, over the data for fiscal year 2009, the 
transition year. Furthermore, as the table shows, the annual numbers 
for both types of inspections have varied over the years, especially 
for hours of service inspections, and there is no indication thus far 
of a change in FRA's emphasis on hours of service. The data for fiscal 
year 2010 are consistent with the statements of some FRA officials, 
who told us FRA placed no special emphasis on hours of service issues 
after RSIA was implemented and has not changed its hours of service 
inspections since the change in the law. According to the officials, 
inspections focus on factors that cause accidents, and hours of 
service issues have caused few, if any, accidents in recent years. 
Most railroads responding to our survey also reported that they did 
not see a change in FRA's handling of hours of service issues. FRA 
did, however, identify hours of service in the National Safety Program 
Plan for fiscal year 2010 as a special-emphasis activity for four of 
FRA's eight regional offices and for the Office of Railroad Safety at 
FRA headquarters. Yet in three of these regional offices, the efforts 
are focused on signal employees rather than T&E employees, the largest 
group of covered employees subject to hours of service limitations. 

Table 2: Hours of Service and Hours of Service Recordkeeping 
Inspections of Freight Railroads, Fiscal Years 2005-2010: 

All inspections; 
2005: 196,594; 
2006: 220,994; 
2007: 239,211; 
2008: 235,492; 
2009: 254,543; 
2010: 269,882. 

Hours of service inspections (percentage of all inspections); 
2005: 264 (0.13%); 
2006: 207 (0.09%); 
2007: 198 (0.08%); 
2008: 309 (0.13%); 
2009: 157 (0.06%); 
2010: 250 (0.09%). 

Recordkeeping inspections (percentage of all inspections); 
2005: 962 (0.49%); 
2006: 1,295 (0.59%); 
2007: 1,064 (0.44%); 
2008: 1,056 (0.45%); 
2009: 1,008 (0.40%); 
2010: 1,016 (0.38%). 

Source: GAO analysis of FRA data. 

[End of table] 

Overall, from fiscal year 2005 through fiscal year 2010, hours of 
service and hours of service recordkeeping inspections accounted for a 
very small percentage of FRA inspections of freight railroads--less 
than 1 percent of all FRA inspections conducted on freight railroads 
each year during this period, as indicated in table 2. Furthermore, 
although operating practices inspectors conducted about 83 percent of 
the hours of service and about 79 percent of the hours of service 
recordkeeping inspections, these inspections accounted for less than 3 
percent of all operating practices inspections conducted at freight 
railroads during fiscal years 2005 through 2010 (see figure 10). 

Figure 10: Proportion of Hours of Service and Hours of Service 
Recordkeeping Inspections Conducted by Selected FRA Disciplines, and 
As a Portion of All Operating Practices, Fiscal Years 2005-2010: 

[Refer to PDF for image: pie-chart and sub-chart] 

Operating practices inspections: 
Hours of service inspections by inspector discipline (percentage): 
* Operating practices: 83.1%; 
* Signal:14.3%; 
* Other: 2.6%; 
Total: 2.0%. 

Hours of service recordkeeping inspections by inspector discipline 
(percentage): 
* Operating practices: 78.6%; 
* Signal: 16.4%; 
* Other: 5.0%; 
Total: 0.5%. 

All other inspection activities: 97.5%. 

Source: GAO analysis of FRA data. 

[End of figure] 

While our analysis does not indicate any notable change in FRA's 
overall emphasis on compliance with hours of service and hours of 
service recordkeeping requirements, it may show proportionally greater 
attention to the class I railroads, especially for hours of service 
(see figure 11). As previously noted, class I railroads account for 
over two-thirds of the total rail mileage operated in the United 
States. For both hours of service and hours of service recordkeeping, 
the number of inspections increased for the class I railroads and 
decreased for the class II and III railroads from fiscal year 2009 to 
fiscal year 2010. Again, however, the data are for a single year, and 
it is unclear whether any observed change will persist. According to 
FRA officials, there are no plans to require additional hours of 
service or hours of service recordkeeping inspections unless there is 
evidence of an increase in noncompliance on the part of the railroads, 
or there is an increase in complaints about violations of the hours of 
service laws. 

Figure 11: Hours of Service and Hours of Service Recordkeeping 
Inspections of Freight Railroads, Fiscal Years 2005-2010: 

[Refer to PDF for image: multiple line graph] 

Number of inspections: 

Fiscal year: 2005; 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 6; 
Class II/III hours of service recordkeeping inspections: 13; 
Class I hours of service recordkeeping inspections: 51. 

Class II/III hours of service inspections: 3; 
Class I hours of service inspections: 18; 
Class II/III hours of service recordkeeping inspections: 27; 
Class I hours of service recordkeeping inspections: 71. 

Class II/III hours of service inspections: 8; 
Class I hours of service inspections: 14; 
Class II/III hours of service recordkeeping inspections: 22; 
Class I hours of service recordkeeping inspections: 52. 

Class II/III hours of service inspections: 3; 
Class I hours of service inspections: 24; 
Class II/III hours of service recordkeeping inspections: 35; 
Class I hours of service recordkeeping inspections: 82. 

Class II/III hours of service inspections: 6; 
Class I hours of service inspections: 17; 
Class II/III hours of service recordkeeping inspections: 33; 
Class I hours of service recordkeeping inspections: 49. 

Class II/III hours of service inspections: 6; 
Class I hours of service inspections: 49; 
Class II/III hours of service recordkeeping inspections: 36; 
Class I hours of service recordkeeping inspections: 72. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 13; 
Class II/III hours of service recordkeeping inspections: 22; 
Class I hours of service recordkeeping inspections: 30. 

Class II/III hours of service inspections: 0; 
Class I hours of service inspections: 16; 
Class II/III hours of service recordkeeping inspections: 15; 
Class I hours of service recordkeeping inspections: 62. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 16; 
Class II/III hours of service recordkeeping inspections: 26; 
Class I hours of service recordkeeping inspections: 57. 

Class II/III hours of service inspections: 8; 
Class I hours of service inspections: 20; 
Class II/III hours of service recordkeeping inspections: 21; 
Class I hours of service recordkeeping inspections: 34. 

Class II/III hours of service inspections: 6; 
Class I hours of service inspections: 18; 
Class II/III hours of service recordkeeping inspections: 42; 
Class I hours of service recordkeeping inspections: 45. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 7; 
Class II/III hours of service recordkeeping inspections: 20; 
Class I hours of service recordkeeping inspections: 45. 

Fiscal year: 2006;  

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 4; 
Class II/III hours of service recordkeeping inspections: 15; 
Class I hours of service recordkeeping inspections: 49. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 7; 
Class II/III hours of service recordkeeping inspections: 33; 
Class I hours of service recordkeeping inspections: 78. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 14; 
Class II/III hours of service recordkeeping inspections: 29; 
Class I hours of service recordkeeping inspections: 72. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 23; 
Class II/III hours of service recordkeeping inspections: 34; 
Class I hours of service recordkeeping inspections: 123. 

Class II/III hours of service inspections: 4; 
Class I hours of service inspections: 25; 
Class II/III hours of service recordkeeping inspections: 24; 
Class I hours of service recordkeeping inspections: 117. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 28; 
Class I hours of service recordkeeping inspections: 81. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 21; 
Class I hours of service recordkeeping inspections: 51. 

Class II/III hours of service inspections: 5; 
Class I hours of service inspections: 16; 
Class II/III hours of service recordkeeping inspections: 30; 
Class I hours of service recordkeeping inspections: 98. 

Class II/III hours of service inspections: 8; 
Class I hours of service inspections: 8; 
Class II/III hours of service recordkeeping inspections: 32; 
Class I hours of service recordkeeping inspections: 87. 

Class II/III hours of service inspections: 6; 
Class I hours of service inspections: 8; 
Class II/III hours of service recordkeeping inspections: 27; 
Class I hours of service recordkeeping inspections: 38. 

Class II/III hours of service inspections: 3; 
Class I hours of service inspections: 8; 
Class II/III hours of service recordkeeping inspections: 26; 
Class I hours of service recordkeeping inspections: 43. 

Class II/III hours of service inspections: 26; 
Class I hours of service inspections: 11; 
Class II/III hours of service recordkeeping inspections: 61; 
Class I hours of service recordkeeping inspections: 98. 

Fiscal year: 2007; 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 15; 
Class II/III hours of service recordkeeping inspections: 29; 
Class I hours of service recordkeeping inspections: 34. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 18; 
Class II/III hours of service recordkeeping inspections: 31; 
Class I hours of service recordkeeping inspections: 58. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 35; 
Class I hours of service recordkeeping inspections: 33. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 13; 
Class II/III hours of service recordkeeping inspections: 29; 
Class I hours of service recordkeeping inspections: 78. 

Class II/III hours of service inspections: 4; 
Class I hours of service inspections: 29; 
Class II/III hours of service recordkeeping inspections: 32; 
Class I hours of service recordkeeping inspections: 74. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 16; 
Class II/III hours of service recordkeeping inspections: 38; 
Class I hours of service recordkeeping inspections: 66. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 32; 
Class I hours of service recordkeeping inspections: 67. 

Class II/III hours of service inspections: 6; 
Class I hours of service inspections: 9; 
Class II/III hours of service recordkeeping inspections: 31; 
Class I hours of service recordkeeping inspections: 70. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 14; 
Class II/III hours of service recordkeeping inspections: 20; 
Class I hours of service recordkeeping inspections: 60. 

Class II/III hours of service inspections: 3; 
Class I hours of service inspections: 9; 
Class II/III hours of service recordkeeping inspections: 30; 
Class I hours of service recordkeeping inspections: 48. 

Class II/III hours of service inspections: 4; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 18; 
Class I hours of service recordkeeping inspections: 74. 

Class II/III hours of service inspections: 5; 
Class I hours of service inspections: 11; 
Class II/III hours of service recordkeeping inspections: 20; 
Class I hours of service recordkeeping inspections: 57. 

Fiscal year: 2008; 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 29; 
Class I hours of service recordkeeping inspections: 44. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 32; 
Class I hours of service recordkeeping inspections: 40. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 3; 
Class II/III hours of service recordkeeping inspections: 11; 
Class I hours of service recordkeeping inspections: 31. 

Class II/III hours of service inspections: 12; 
Class I hours of service inspections: 17; 
Class II/III hours of service recordkeeping inspections: 24; 
Class I hours of service recordkeeping inspections: 87. 

Class II/III hours of service inspections: 63; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 43; 
Class I hours of service recordkeeping inspections: 73. 

Class II/III hours of service inspections: 10; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 40; 
Class I hours of service recordkeeping inspections: 64. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 18; 
Class II/III hours of service recordkeeping inspections: 37; 
Class I hours of service recordkeeping inspections: 43. 

Class II/III hours of service inspections: 0; 
Class I hours of service inspections: 11; 
Class II/III hours of service recordkeeping inspections: 20; 
Class I hours of service recordkeeping inspections: 47. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 13; 
Class II/III hours of service recordkeeping inspections: 28; 
Class I hours of service recordkeeping inspections: 123. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 81; 
Class II/III hours of service recordkeeping inspections: 27; 
Class I hours of service recordkeeping inspections: 52. 

Class II/III hours of service inspections: 0; 
Class I hours of service inspections: 17; 
Class II/III hours of service recordkeeping inspections: 24; 
Class I hours of service recordkeeping inspections: 55. 

Class II/III hours of service inspections: 4; 
Class I hours of service inspections: 12; 
Class II/III hours of service recordkeeping inspections: 28; 
Class I hours of service recordkeeping inspections: 54. 

Fiscal year: 2009; 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 7; 
Class II/III hours of service recordkeeping inspections: 34; 
Class I hours of service recordkeeping inspections: 60. 

Class II/III hours of service inspections: 12; 
Class I hours of service inspections: 14; 
Class II/III hours of service recordkeeping inspections: 39; 
Class I hours of service recordkeeping inspections: 36. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 11; 
Class II/III hours of service recordkeeping inspections: 16; 
Class I hours of service recordkeeping inspections: 62. 

Class II/III hours of service inspections: 0; 
Class I hours of service inspections: 9; 
Class II/III hours of service recordkeeping inspections: 39; 
Class I hours of service recordkeeping inspections: 73. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 12; 
Class II/III hours of service recordkeeping inspections: 20; 
Class I hours of service recordkeeping inspections: 61. 

Class II/III hours of service inspections: 28; 
Class I hours of service inspections: 3; 
Class II/III hours of service recordkeeping inspections: 51; 
Class I hours of service recordkeeping inspections: 53. 

Class II/III hours of service inspections: 7; 
Class I hours of service inspections: 9; 
Class II/III hours of service recordkeeping inspections: 40; 
Class I hours of service recordkeeping inspections: 65. 

Class II/III hours of service inspections: 0; 
Class I hours of service inspections: 9; 
Class II/III hours of service recordkeeping inspections: 22; 
Class I hours of service recordkeeping inspections: 46. 

Class II/III hours of service inspections: 0; 
Class I hours of service inspections: 2; 
Class II/III hours of service recordkeeping inspections: 30; 
Class I hours of service recordkeeping inspections: 35. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 2; 
Class II/III hours of service recordkeeping inspections: 32; 
Class I hours of service recordkeeping inspections: 44. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 8; 
Class II/III hours of service recordkeeping inspections: 22; 
Class I hours of service recordkeeping inspections: 48. 

Class II/III hours of service inspections: 3; 
Class I hours of service inspections: 15; 
Class II/III hours of service recordkeeping inspections: 28; 
Class I hours of service recordkeeping inspections: 52. 

Fiscal year: 2010; 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 20; 
Class I hours of service recordkeeping inspections: 45. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 6; 
Class II/III hours of service recordkeeping inspections: 29; 
Class I hours of service recordkeeping inspections: 44. 

Class II/III hours of service inspections: 6; 
Class I hours of service inspections: 9; 
Class II/III hours of service recordkeeping inspections: 28; 
Class I hours of service recordkeeping inspections: 54. 

Class II/III hours of service inspections: 2; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 44; 
Class I hours of service recordkeeping inspections: 71. 

Class II/III hours of service inspections: 4; 
Class I hours of service inspections: 17; 
Class II/III hours of service recordkeeping inspections: 36; 
Class I hours of service recordkeeping inspections: 62. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 64; 
Class II/III hours of service recordkeeping inspections: 29; 
Class I hours of service recordkeeping inspections: 70. 

Class II/III hours of service inspections: 6; 
Class I hours of service inspections: 27; 
Class II/III hours of service recordkeeping inspections: 25; 
Class I hours of service recordkeeping inspections: 64. 

Class II/III hours of service inspections: 3; 
Class I hours of service inspections: 19; 
Class II/III hours of service recordkeeping inspections: 22; 
Class I hours of service recordkeeping inspections: 95. 

Class II/III hours of service inspections: 1; 
Class I hours of service inspections: 10; 
Class II/III hours of service recordkeeping inspections: 27; 
Class I hours of service recordkeeping inspections: 80. 

Class II/III hours of service inspections: 0; 
Class I hours of service inspections: 12; 
Class II/III hours of service recordkeeping inspections: 16; 
Class I hours of service recordkeeping inspections: 47. 

Class II/III hours of service inspections: 3; 
Class I hours of service inspections: 28; 
Class II/III hours of service recordkeeping inspections: 16; 
Class I hours of service recordkeeping inspections: 44. 

Source: GAO analysis of FRA data. 

[End of figure] 

With just one full year's worth of data since RSIA took effect, we 
could not discern any changes in FRA's hours of service enforcement 
priorities. One indicator--the portion of defects identified during 
inspections that resulted in violations being processed for 
enforcement--showed no consistent direction, fluctuating variously up 
or down in fiscal year 2009 from fiscal year 2008 and then reversing 
direction the following year. Another indicator--enforcement actions 
taken--also fluctuated, with the number of hours of service 
enforcement actions going up in fiscal year 2009 from fiscal year 
2008, and then dropping again in fiscal year 2010. Meanwhile, hours of 
service recordkeeping enforcement actions took the opposite path 
during the same period, first dropping and then rising (see table 3). 

Table 3: Hours of Service and Hours of Service Recordkeeping 
Enforcement Actions Taken against Freight Railroads, Fiscal Years 2005-
2010: 

Hours of service: Enforcement actions; : 
2005: 41; 
2006: 51; 
2007: 54; 
2008: 45; 
2009: 51; 
2010: 33. 

Hours of service: Total proposed penalties[A]; 
2005: $167,433; 
2006: $207,525; 
2007: $163,478; 
2008: $181,469; 
2009: $175,016; 
2010: $67,670. 

Hours of service recordkeeping: Enforcement actions; 
2005: 27; 
2006: 47; 
2007: 67; 
2008: 63; 
2009: 41; 
2010: 47. 

Hours of service recordkeeping: Total proposed penalties; 
2005: $84,278; 
2006: $178,189; 
2007: $198,283; 
2008: $156,930; 
2009: $100,736; 
2010: $167,660. 

Source: GAO analysis of FRA enforcement data. 

[A] Penalty figures are shown in constant fiscal year 2011 dollars. 

[End of table] 

RSIA Provisions for Pilot Projects and Waivers Create Opportunities to 
Analyze Safety Effects of Approved Alternatives: 

Besides establishing new hours of service requirements, RSIA provided 
for pilot projects and waivers of compliance with hours of service 
requirements, both of which would create opportunities for FRA and 
railroads to analyze the effects on safety of approved alternatives to 
the new hours of service requirements. FRA has been unable to 
implement two pilot projects mandated under RSIA because no railroads 
have chosen to participate, and has not exercised its pre-RSIA 
authority to approve voluntary pilot projects designed to examine the 
fatigue-reduction potential of alternatives to the current hours of 
service laws because of flaws in the applications it received. FRA 
also has the authority to approve petitions for waivers of hours of 
service requirements in certain circumstances and has approved waiver 
petitions for some railroads. RSIA required FRA to conduct the 
mandated pilot projects by October 2010, and to report on the 
voluntary pilot projects no later than December 31, 2012;[Footnote 55] 
however, RSIA does not require FRA to analyze or report on the safety 
effects of approved waiver petitions, and FRA has not taken steps to 
do so. 

Mandated and Voluntary Pilot Projects: 

RSIA required FRA to conduct at least two pilot projects of sufficient 
size and scope to analyze specific practices that could be used to 
reduce fatigue for T&E and other railroad employees covered by hours 
of service requirements. The first pilot project called for the 
railroad to give a covered employee at least 10 hours' advance notice 
of a shift assignment. Advance notice of 2 to 4 hours is typical in 
the industry today. The second pilot project would have created 
defined shifts for covered employees who receive unscheduled shift 
calls, such that those employees would be subject to call every other 
shift, instead of at any time. FRA has not been able to implement 
either of these mandated pilot projects because no railroad has 
expressed interest in implementing either project. According to FRA 
officials, the agency lacks authority to compel railroads to 
participate. According to both FRA and railroad officials, railroads 
have not chosen to participate in the pilot projects mandated in the 
legislation because doing so could put a participating railroad at a 
competitive disadvantage. More specifically, both projects would 
decrease a railroad's flexibility to assign covered train employees to 
report as circumstances warrant--in the first case by requiring 
advance notice of at least 10 hours, rather than the typical practice 
of 2 to 4 hours notice, and in the second case by reducing the pool of 
employees on call by half. Because freight railroads try to work to 
accommodate their customers, often with last-minute scheduling 
changes, it is important for them to remain flexible so they can 
compete with other railroads and other modes of transportation, such 
as trucks. 

While FRA was unable to conduct the two pilot projects mandated in 
RSIA, it still has authority to approve voluntary pilot projects. This 
authority, which predates RSIA, allows FRA to approve joint petitions 
from railroads and nonprofit employee labor organizations representing 
directly affected covered service employees of the railroads for 
waivers of compliance with the hours of service law in order to 
demonstrate the possible benefits of implementing alternatives to 
strict adherence to the law, including requirements for maximum on-
duty and minimum off-duty periods.[Footnote 56] According to FRA 
officials, there was little interest in obtaining waivers for pilot 
projects prior to the passage of RSIA. Since May 2009, however, FRA 
has received five petitions for waivers of compliance with hours of 
service requirements in order to implement voluntary pilot projects. 
FRA dismissed two of these petitions, because they were not filed 
jointly, as required, by a railroad and the employee labor 
organizations representing the affected employees. FRA approved two 
other petitions for pilot projects requesting waivers, but both were 
designed to provide administrative alternatives rather than 
alternatives to the requirements concerning maximum on-duty and 
minimum off-duty periods.[Footnote 57] In approving these petitions, 
FRA noted that because the proposed pilot projects were administrative 
in nature, they would not impinge on the likely performance or safety 
of the railroads. Finally, FRA rejected one petition that was designed 
to identify alternatives to the hours of service laws for addressing 
fatigue. This petition, filed by the ASLRRA on behalf of its members, 
sought approval for a pilot project that would, among other things, 
develop and identify alternative methods to mitigate the risk of 
fatigue without strict adherence to the new hours of service 
requirements. While acknowledging that ASLRRA raised salient issues 
for short-line and small railroads, FRA rejected the petition, noting 
that it lacked a thorough explanation of the conditions and controls 
under which the pilot project would be operated to ensure the safety 
of railroad operations and participating employees. Moreover, 
according to FRA, the petition failed to identify what additional 
relief from the hours of service laws was necessary to implement the 
pilot project. 

Other than the petition filed by ASLRRA, FRA has received no petitions 
for waivers of compliance with hours of service requirements in order 
to implement voluntary pilot projects that could demonstrate the 
fatigue-reduction potential of alternatives to RSIA provisions. 
Information gathered from railroads operating monitored pilot projects 
could be analyzed to assess the effectiveness of specific practices 
being used to reduce fatigue. Such information could also be used to 
examine the effects on safety of, for example, increasing rest 
requirements for some shifts that extend into night hours, providing 
fewer hours of rest for employees resting away from their home 
terminal, or decreasing rest requirements for those covered employees 
working only regular daytime shifts. The results of such analysis 
could be used to inform the RSIA-required report to Congress by 
December 2012 on the effectiveness of the voluntary pilot projects. 

Waivers of Hours of Service Requirements: 

Even though no pilot projects currently afford opportunities for 
gathering and analyzing data on the safety effects of alternatives to 
the new hours of service requirements, FRA could obtain such data from 
railroads operating with approved waivers of compliance with the new 
hours of service requirements. FRA has the authority to approve 
petitions for waivers of the statutory requirements related to the 
consecutive work day limits if a collective bargaining agreement 
provides for a different arrangement, and such an arrangement is in 
the public interest and consistent with railroad safety.[Footnote 58] 
As of June 30, 2011, FRA had received 17 petitions for waivers of 
compliance with hours of service requirements and had fully approved 8 
of them, including 1 filed by ASLRRA that covers 142 of its member 
railroads.[Footnote 59] In total, 157 railroads have approved waivers 
of compliance with hours of service requirements, 2 of which are class 
I railroads. The remainder are class II or III railroads. The approved 
waivers recognize that the risk of fatigue is greater for night shifts 
than for day shifts, as discussed earlier in this report. 
Specifically, all of the approved waivers allow scheduled shifts of 6 
consecutive work days followed by 24 hours' rest (6 by 1 schedules), 
rather than the 48 hours' rest (6 by 2 schedules) required by law, 
provided that the shifts during those 6 consecutive days do not extend 
into the hours between midnight and 6 a.m. Table 4 provides 
information on the disposition of the waiver petitions submitted to 
FRA from May 2009 through June 2011. 

Table 4: Disposition of Petitions for Waivers of Limitations on Hours 
of Service for Railroad Employees, May 2009 through June 2011: 

Disposition of petition (number of petitions): Approved (8); 
Description: 24 hours' rest (rather than 48 hours as required in law) 
following 6 consecutive days of shift starts, provided that those 6 
days do not extend into the hours between midnight and 6 a.m. 

Disposition of petition (number of petitions): Partially approved (1); 
Description: Approved 24 hours' rest (rather than 48 hours as required 
in law) following 6 consecutive days of shift starts, provided that 
those 6 days do not extend into the hours between midnight and 6 a.m. 
However, denied a waiver for on-call employees on a pattern of 11 days 
on call and 3 days off. 

Disposition of petition (number of petitions): Denied (1); 
Description: Waiver of the 276-hour total monthly work hour cap for 
management employees who engage in limited covered service for no more 
than 25 percent of their hours during a month. 

Disposition of petition (number of petitions): Dismissed (3); 
Description: Petitions did not meet the requirements for a joint 
filing by a railroad and the labor organization representing the 
directly affected employees (or for employee concurrence where there 
is no labor organization representation). In addition, one petition 
requested a waiver from a provision that FRA does not have the 
authority to waive. 

Disposition of petition (number of petitions): Withdrawn (2); 
Description: One railroad withdrew to be included in the ASLRRA 
petition, while the other withdrew its petition and sought no other 
relief. 

Disposition of petition (number of petitions): Pending decisions (2); 
Description: Both petitions request approval for 24 hours' rest 
(rather than 48 hours as required in law) following 6 consecutive days 
of shift starts. 

Source: GAO analysis of FRA data. 

[End of table] 

RSIA did not require FRA to collect data or report on the safety 
effects of approved waiver petitions, as it did for the voluntary 
pilot projects, and FRA has not taken steps to do so. According to an 
FRA official, establishing a level of fatigue among employees working 
under the conditions of one of the approved voluntary waivers would 
require an evaluation of the employees' work and rest schedules using 
a fatigue model such as FAST. The easiest way to collect such data, 
the official said, would be to have inspectors evaluate these employee 
schedules at randomly selected railroads. The official acknowledged 
that having data about railroads operating under waivers could help 
determine the feasibility of alternatives to RSIA's current 
requirements, such as a modification of the requirement for 48 hours' 
rest after 6 consecutive work days for certain scheduled shifts. 

Conclusions: 

Since RSIA was only recently implemented, it is still too early to 
determine whether its changes to hours of service requirements will 
materially affect freight railroad safety. Initial indications are 
positive, as our analysis of selected covered T&E employee work 
schedules shows. Rest time for these employees has increased, and the 
amount of time they work at a high risk of fatigue has decreased--up 
to about 36 percent for some railroads. However, as can be expected of 
changes in laws to improve safety, these benefits have also resulted 
in some costs to both railroad employees and the industry. Our work 
shows that some covered employees saw reductions in their work hours 
and, according to information from our survey of the railroad industry 
and related interviews, many railroads made changes in crew and train 
schedules, incurred additional costs to hire new employees or bring 
employees back from furlough to maintain operations and comply with 
the law, and saw reductions in their ability to provide service to 
customers when and where needed. More important, although the time 
spent working at a high risk of fatigue decreased for some T&E 
employees, RSIA did not address work performed during night hours, 
which, according to both scientific literature and our analysis of 
covered T&E employee work schedules, represents a major factor in 
fatigue risk. Therefore, opportunities for reducing the risk of 
fatigue remain, especially since night work is integral to freight 
rail operations. Moreover, we believe further analysis of the safety 
implications of both day and night work, and of actions that could be 
taken to mitigate the associated fatigue risks, could point to 
opportunities for trade-offs that would reduce the overall risk of 
fatigue yet potentially allow for a relaxation of RSIA provisions that 
railroads and employees said were particularly burdensome to them--
such as the consecutive work day limits before mandatory rest. 

The federal government also plays an important role in helping promote 
safe railroad operations through its inspection and enforcement 
actions. FRA's risk-based approach to oversight is intended to align 
the agency's inspection and enforcement resources with risks. The NIP 
provides a good foundation for doing this, including the use of local 
input to ensure resources are focused on the specific risks that may 
lead to accidents. As we saw from the data, it is too soon to 
determine if the emphasis FRA has so far given to hours of service 
requirements best aligns with the risks associated with the RSIA 
changes and this will bear watching going forward. Additionally, in 
our view, FRA is missing opportunities to better identify the 
potential costs, benefits, and safety implications of alternatives to 
the current hours of service requirements. While voluntary pilot 
projects were envisioned in RSIA and offer the opportunity for FRA and 
railroads to try alternative approaches and learn from them, interest 
from the industry has, to date, been low. Realizing the full benefits 
from pilot projects will require additional outreach to the rail 
industry and other stakeholders to generate ideas on how pilot 
projects could be structured so they generate interest and 
participation, including ways to minimize potential competitive 
disadvantages to participants. Both pilot projects and waivers could 
generate information that would be of use in aligning oversight 
resources with risks, analyzing fatigue issues, and deciding how to 
reduce fatigue risks in the railroad industry, as well as informing 
FRA's December 2012 report to Congress on voluntary pilot projects. 

Recommendations for Executive Action: 

To ensure that FRA's implementation of hours of service requirements 
in the freight railroad industry maximizes opportunities to reduce the 
risks of accidents and incidents related to fatigue, we recommend that 
the Secretary of Transportation direct the Administrator of FRA to 
take the following action: 

* Evaluate and develop recommendations about the relative impact of 
consecutive days worked and work performed during night hours on the 
potential for fatigue and risk of accidents in the freight railroad 
industry. This evaluation should attempt to determine if taking night 
work into consideration in the hours of service limitations (such as 
by requiring more rest after night work) would enable some relaxation 
of the current limits on consecutive days worked before rest is 
required in such a way that the same or better overall reduction in 
fatigue risk occurs while mitigating negative effects on employees and 
railroad operations. In performing this evaluation, FRA should 
consider scientific and medical research related to fatigue and 
fatigue abatement and data from pilot projects and waivers of 
compliance with hours of service requirements that relate to fatigue 
levels and consecutive days worked and work performed at night. FRA 
should also communicate the results of the evaluation to appropriate 
congressional committees for their consideration. 

To improve FRA's targeting of its inspection resources and 
understanding of the effect of work hours on fatigue, we recommend 
that the Secretary of Transportation direct the Administrator of FRA 
to take the following action: 

* Work with the railroad industry to identify pilot projects that 
could be implemented to test the fatigue reduction potential of 
alternatives to the current hours of service laws. Also, collect 
safety indicator and accident and incident data from participants in 
pilot projects and railroads with waivers of compliance with hours of 
service requirements to determine the effects of such pilot projects 
and waivers on covered employee fatigue and participant safety 
performance. FRA should then incorporate the results of both efforts 
into the risk assessment process used to determine the allocation of 
inspection resources and report the results to appropriate committees 
of Congress. 

Agency Comments and Our Evaluation: 

We provided a draft of this report and the e-supplement to DOT for 
review and comment. We met with FRA officials, including the Deputy 
Chief Counsel, on September 19, 2011. DOT expressed concerns about a 
portion of our second recommendation that it incorporate activity-
level data into the NIP's development of inspection priorities and 
that it add a code for hours of service issues to the Accident and 
Incident Reporting System. According to DOT, the NIP provides a 
comprehensive framework to manage hundreds of competing inspection 
activities--including hours of service inspections--and incorporating 
activity-level data as we suggested would imply a level of precision 
that does not exist. DOT also emphasized the value of FRA inspectors' 
input into the priority-setting process and suggested that an 
increased reliance on data could reduce FRA's flexibility and 
efficiency in responding to and managing local issues. In addition, 
DOT considered adding an hours of service code to the Accident and 
Incident Reporting System redundant, since railroads are already 
required to report excess service hours to FRA every month. 
Furthermore, FRA said that adding such a code would not be helpful, 
since an hours of service violation may not indicate fatigue. 
According to FRA officials, a covered employee could be fatigued while 
complying with hours of service requirements or a covered employee 
could be noncompliant with hours of service requirements without being 
fatigued. FRA officials told us that accidents and incidents generally 
occur because someone misaligned a switch, failed to observe a signal, 
or failed to take some other physical action. These may or may not 
have been caused by fatigue but adding a code for hours of service 
would not indicate fatigue levels. FRA officials noted that several 
cause codes in the Accident and Incident Reporting System can indicate 
fatigue and that FRA investigators follow up to assess the role of 
fatigue when railroads identify those codes as causes of accidents. 
FRA officials also noted that a review of recent reports on rail 
accidents, including reports from the National Transportation Safety 
Board, found none that identified hours of service as a cause of an 
accident. 

After we met with FRA officials, they provided additional information 
on how FRA uses activity-based inspection data (including hours of 
service data) to develop the NIP and furnished us with a list of codes 
included in the Accident and Incident Reporting System that show some 
correlation with fatigue. In light of our discussions with FRA and our 
analysis of the information it subsequently provided, we withdrew the 
portions of our second recommendation that FRA incorporate activity- 
level data into the risk assessment process and add one or more codes 
to the Accident and Incident Reporting System to identify the role of 
hours of service in railroad accidents. The information provided by 
FRA shows that hours of service activity-based data is being used to 
develop the NIP and adding one or more codes to the Accident and 
Incident Reporting System for hours of service might not be helpful in 
identifying broader issues of the role fatigue plays in accidents. 
Rather, such information is more likely to come from FRA's accident 
investigations, which can also identify if violations of hours of 
service requirements play a role in rail accidents. 

FRA officials also raised concerns about the wording of some 
definitions used in our survey of the rail industry about hours of 
service issues. FRA questioned whether the rail industry was familiar 
enough with requirements of the law so that the definitions we used 
did not result in inaccurate responses. We did not change the wording 
of the definitions contained in the survey presented in our e-
supplement to this report as a result of FRA's comments because the e-
supplement is meant to present the survey as it was made available to 
respondents. We believe the information provided by the survey is 
accurate and respondents understood our survey and RSIA requirements 
sufficiently to provide appropriate responses. To this end, we fully 
pretested the survey prior to administering it; pretest participants 
raised no substantive concerns about the terms defined in the survey. 
Our e-supplement product (GAO-11-894SP) contains additional 
information about FRA's comments. 

DOT also provided technical comments that we incorporated as 
appropriate. 

As agreed with your offices, unless you publicly announce the contents 
of this report earlier, we plan no further distribution of this report 
until 30 days from the report date. At that time, we will send copies 
of this report to the appropriate congressional committees, the 
Secretary of Transportation, the Administrator of FRA, and the 
Director of the Office of Management and Budget. The report will also 
be available at no charge on the GAO website at [hyperlink, 
http://www.gao.gov]. 

If you or your staff have any questions about this report, please 
contact me at (202) 512-2834 or flemings@gao.gov. Contact points for 
our Offices of Congressional Relations and Public Affairs may be found 
on the last page of this report. GAO staff who made major 
contributions to this report are listed in appendix III. 

Signed by: 

Susan A. Fleming: 
Director, Physical Infrastructure Team: 

[End of section] 

Appendix I: Objectives, Scope, and Methodology: 

To better understand the changes to freight railroad hours of service 
requirements made by the Rail Safety Improvement Act of 2008 (RSIA), 
we reviewed the (1) impacts of the hours of service changes on the 
covered train and engine (T&E) workforce, including potential impacts 
on fatigue; (2) operational and administrative impacts of the hours of 
service changes on the railroad industry; and (3) actions taken by the 
Federal Railroad Administration (FRA) to oversee compliance with hours 
of service requirements and implement RSIA provisions related to hours 
of service pilot projects and waivers. 

The scope of this engagement was limited to the freight railroad 
industry. The RSIA hours of service requirements became effective for 
these railroads on July 16, 2009. We did not include commuter and 
intercity passenger railroads, since at the time of our work FRA was 
in the process of developing new hours of service requirements for 
these railroads. Our scope included freight railroads of all sizes. 
The freight railroad industry is divided into three classes: I, II, 
and III, based on their operating revenues. In 2009, annual operating 
revenues were at least $378.8 million for class I railroads, between 
$30.3 million and $378.8 million for class II railroads, and less than 
$30.3 million for class III railroads. The class designation differs 
slightly from another designation that FRA uses for accident and 
incident reporting, under which railroads are divided into groups. 
FRA's group 1 is equivalent to class I. The division between groups 2 
and 3 is based on the total number of annual work hours reported to 
FRA. Group 2 railroads report 400,000 or more total annual work hours 
but are not class I railroads, and group 3 railroads report less than 
400,000 total annual work hours. According to FRA officials, groups 2 
and 3 are not necessarily the same as classes II and III, but the 
differences may not be large. For reporting purposes we use the class 
designation because (1) it is a common means of identifying railroads 
and (2) the railroads included in class II or III may not be 
significantly different from those in group 2 or 3, respectively. 

The following describes some of the key methodologies we used to 
address our objectives. 

Structured Collection and Analysis of Covered Employee Work Schedules 
and Use of Fatigue Models: 

To address the impacts of the hours of service changes on the covered 
T&E workforce, including potential impacts on fatigue, we collected 
and analyzed covered T&E employee work schedule data and used the 
Fatigue Audit InterDyneTM (FAID) biomathematical fatigue model. We 
initially requested covered T&E employee work schedules for all 7 
class I, 15 class II, and a sample of 86 class III railroads. However, 
not all the class II and III railroads we contacted had electronic 
records; instead, most maintained paper-based hours of service 
records. We determined that the process for collecting and 
transcribing the paper-based work schedules into electronic format for 
analysis was not feasible given our time and resource constraints. 
Accordingly, we focused our data collection on electronic records from 
class I railroads and those class II railroads that responded to our 
inquiries and could provide electronic hours of service records. In 
addition, we conducted focused telephone interviews with 69 randomly 
sampled class III railroads with 5 or more full-time-equivalent 
employees covered by hours of service requirements to obtain 
information about their operations. 

For our analysis of electronic hours of service records, we included 
all work schedules for all covered T&E employees that had work 
schedule data for both May 2008 and May 2010 and had at least 7 days 
of scheduled work in both these months. All 7 class I railroads 
submitted the requested records, and 6 class II railroads[Footnote 60] 
provided electronic records that met our requirements. The final data 
set covers the May 2008 and May 2010 work records for 52,205 class I 
covered T&E employees and 963 class II covered T&E employees. We 
selected May 2008 and May 2010 for our analysis because they represent 
months before and after RSIA's implementation. In addition, choosing 
the same month for both years helps to avoid any seasonal differences 
in the rail industry. We also discussed the time frames for our 
analysis with rail industry representatives, and they generally agreed 
with our selection. To assess the reliability of the data provided, we 
performed tests to detect and eliminate anomalies such as duplicate 
records, overlapping shifts, shifts with start or end time errors, and 
data for employees who did not work in both time periods. Where 
appropriate, we contacted railroads to correct these anomalies. We 
also sent a questionnaire to the railroads to obtain information about 
the quality control procedures for their electronic systems. We 
determined that the data were sufficiently reliable for the purposes 
of this report. 

Our analysis of employee work schedules was focused on answering key 
questions designed to identify the effects of RSIA's changes on the 
covered T&E workforce, such as whether total work time changed. We 
analyzed class I and class II work schedules separately because this 
approach allowed us to recognize there may be operational differences 
between the two classes of railroads. Among other things, we examined 
work schedule data to determine the total hours worked, total shifts 
worked, total rest time, and total hours worked at night[Footnote 61] 
in both time periods for both classes of railroad. Total hours worked 
and total shifts worked were measures we used to determine if there 
were impacts on the amount of work performed. We used total hours of 
work and total hours worked at night along with fatigue model outputs 
as measures for estimating the impact of night work on fatigue risk 
levels. 

We estimated fatigue risk levels for work schedule data using the FAID 
model, a biomathematical fatigue model that has been used for fatigue 
analyses of railroad work schedules. FRA has validated FAID, as well 
as the Fatigue Avoidance Scheduling ToolTM (FAST), for use in 
analyzing the railroad employees' fatigue risk levels--the only two 
models that FRA had validated for such use at the time of our work. 
The FAID model is commonly used in the railroad industry for fatigue 
analysis, and FRA has used FAST to conduct fatigue analyses for 
regulatory purposes (such as reviews of petitions for waivers of 
compliance with hours of service requirements). We performed separate 
fatigue model analyses for class I and class II railroads and included 
in our analyses all the work schedules in our final data set for both 
class I and class II employees. 

In conducting our fatigue analyses, primarily using the FAID model, we 
established a tolerance level--that is, a fatigue score that, if 
breached, indicates a potentially unacceptable level of fatigue risk. 
We selected a fatigue score of 70 as the threshold for high risk of 
fatigue, scores of 61 through 69 for elevated risk of fatigue, and 
scores below 60 for acceptable risk of fatigue. Fatigue experts, the 
rail industry, and FRA differ on acceptable fatigue risk score 
thresholds. We selected 70--a conservative score--as our threshold for 
high risk of fatigue partly because FRA, in its FAID validation and 
calibration report,[Footnote 62] said a FAID score greater than 70 
would indicate extreme fatigue and partly because we wanted to err on 
the side of caution in our use of the model for fatigue analysis. 
After our analysis was performed FRA issued its final hours of service 
rule for passenger rail employees that set the high fatigue threshold 
for FAID at 72. Additionally, we used the FAID model as its producer 
and fatigue experts directed, that is, to analyze aggregate-level data 
to determine fatigue risk among the covered workforce or to analyze 
generic examples of work schedules to determine fatigue risk. We did 
not use the fatigue models to determine fatigue risk for individual 
covered employees. 

To better understand the relationship between night work and fatigue 
we examined whether the number of hours employees worked at night was 
correlated with spending time at high risk of fatigue according to the 
outputs of the FAID fatigue model. In particular we calculated the 
correlation between night hours and the incidence of employees 
spending at least 20 percent of their hours at high risk of fatigue. 
We chose the 20 percent of time at high risk to be consistent with 
FRA's commuter and intercity rail final hours of service rule, in 
which a fatigue model indicating that 20 percent or more time spent at 
high risk of fatigue would trigger further mitigation of a rail work 
schedule by railroads and approval of mitigation by FRA. The 
correlation coefficient was 0.53. 

Both the data we collected and the analysis we performed have 
limitations. As we discussed earlier, an economic recession began in 
2008 and it affected the demand for rail services significantly. For 
the months we collected rail workers' schedules, May 2008 and May 
2010, overall rail operations were different, with considerably higher 
levels of overall rail service in 2008. Although our findings on the 
differences in work schedules across these time periods may be, in 
part, a reflection of the differences in the macroeconomic 
environment, we attempted to mitigate that factor in two ways. We 
avoided choosing months during which the demand for rail service was 
rapidly declining. In May 2008, the recession was not greatly 
impacting the rail industry and by May 2010, demand was picking up 
from its lows during late 2008 and 2009. Also, we only used employees 
that worked in both months under the assumption the same employees 
would likely be performing similar work tasks in the two periods. 

Additionally, the fatigue models have limitations. In particular, 
fatigue models are developed around an average person as the base 
point. Fatigue models do not consider situations specific to an 
individual that could influence whether an individual's fatigue levels 
and score are the same as or different from those calculated by the 
model. Finally, the models incorporate assumptions about sleep time 
and sleep quality, since it is not possible to determine how long or 
how well a specific individual sleeps during non-work time. Even with 
these limitations, we determined that the rail data and fatigue model 
results were sufficiently reliable for the purposes of this report. 

Survey of the Railroad Industry: 

To identify the operational and administrative impacts of the RSIA 
hours of service changes on the railroad industry, we conducted a web- 
based survey of railroads. To identify survey participants, we used 
FRA's 2009 Accident and Incident Reporting database.[Footnote 63] In 
general, federal regulations require that all U.S. railroads report 
monthly to FRA on accidents and incidents that occur on their 
railroads.[Footnote 64] Exceptions include such railroads as those 
that operate freight trains only on track inside an installation that 
is not part of the general railroad system of transportation and rail 
mass transit operations in urban areas that are not connected to the 
general railroad system of transportation.[Footnote 65] Reports are to 
be made about accidents or incidents that occurred during a month or 
to indicate that no accidents or incidents occurred. As noted, FRA 
reports accident and incident data for class I (excluding the National 
Railroad Passenger Corporation, or Amtrak), class II, and class III 
railroads. To ensure the reliability of FRA's database for our 
purposes, we (1) reviewed federal regulations to better understand 
which railroads are required to report accident and incident data and 
which railroads might be exempt from reporting, (2) reviewed relevant 
database documentation, including FRA's guidelines for reporting 
accidents and incidents, to understand what data are reported and what 
controls are used to ensure the reported data are accurate and 
reliable, and (3) interviewed FRA officials and reviewed FRA written 
responses to our questions to understand the controls FRA used to 
ensure the data were accurate and reliable. Based on these steps, we 
believe the database was sufficiently reliable for our needs. 

We used FRA's Accident and Incident Reporting database, since it 
contained the most recent data available on U.S. railroads at the time 
we performed our work. We excluded from the database passenger-related 
railroads, tourist and historic railroads that had limited operations, 
railroads exempt from reporting, and newly started railroads that had 
not yet built up a record of accidents and incidents. The universe of 
freight railroads surveyed included all 7 class I railroads, all 15 
class II railroads, and all class III railroads that had five or more 
full-time-equivalent employees (based on work hours reported to FRA) 
covered by hours of service requirements in 2009. We chose five full- 
time-equivalent employees as our threshold to, among other things, 
eliminate railroads that (1) might be too small to have hours of 
service impacts, (2) operate only part of the year (seasonal or 
intermittent operators), and (3) have only a few employees who may be 
used as needed for operations only within a plant or other 
manufacturing facility. The survey selections were designed to include 
participants from all three classes of railroads representing large, 
medium, and small entities. We calculated the percentage of full-time- 
equivalent employees covered by hours of service requirements based on 
discussions with FRA officials and estimates of the percentage of the 
railroad employee population covered by hours of service requirements 
used by FRA in previous rulemakings. Out of the 561 class III 
railroads in the database, we calculated there were 234 with five or 
more full-time-equivalent employees covered by hours of service 
requirements. We determined that two of these railroads were not 
eligible for the survey because one was not a railroad (it was a 
centralized dispatching center for several railroads) and one had 
ceased operations in 2009 and, according to an official from this 
railroad, had no experience with the RSIA hours of service changes. In 
total, we surveyed 254 railroads--7 class I railroads, 15 class II 
railroads, and 232 class III railroads. 

To develop our survey questions, we relied on a comprehensive list of 
questions that we used to interview railroads about hours of service 
issues before we conducted the survey. We identified key issues from 
the railroads' responses to the questions and used these to develop 
the survey questionnaire. In addition, we conducted four pretests of 
the survey, one with a class I railroad, one with a class II railroad, 
and two with two class III railroads. Two pretests were done in-person 
and the other two were done over the telephone. The railroads were 
selected to get a variety of large, medium, and small railroads. 
During the pretests, we obtained feedback on such things as the type 
of questions being asked, the clarity of the questions, and whether 
additional issues should be included. We used this feedback to revise 
the survey instrument, including adding questions to cover additional 
issues and clarifying certain survey questions. After completing the 
survey questions, we sent an e-mail announcement of the survey to the 
256 railroads initially included in our survey (including the 2 that 
we subsequently excluded as ineligible) on January 10, 2011. These 
railroads were notified that the questionnaire was available online 
and were given unique passwords and usernames on January 13, 2011. We 
sent follow-up e-mail messages on February 1, February 16, and March 
16, 2011, to those railroads that had not yet responded. We conducted 
the survey from January 13, 2011, to April 15, 2011. 

Because we did not survey a sample of railroads, our survey has no 
sampling errors. However, the practical difficulties of conducting any 
survey may introduce nonsampling errors. For example, difficulties in 
interpreting a particular question, or the type of information 
available to some respondents but not others, could introduce unwanted 
variability into the survey results. We took steps in the data 
collection and data analysis stages to minimize such nonsampling 
errors. As we previously indicated, we collaborated with GAO survey 
specialists to design a draft questionnaire and pretested versions of 
the questionnaire with four members of the survey population. From 
these pretests, we made revisions as necessary to reduce the 
likelihood of nonresponse and reporting errors on our questions. We 
examined the survey results and performed computer analyses to 
identify inconsistencies and other indications of error and addressed 
such issues, where possible. A second, independent analyst checked the 
accuracy of all computer analyses to minimize the likelihood of errors 
in data processing. In addition, GAO analysts answered respondents' 
questions and resolved difficulties that respondents had in answering 
our questions. 

The overall response rate for this survey was 72 percent, with 7 out 
of 7 of class I railroads, 14 out of 15 class II railroads, and 163 
out of 232 class III railroads responding. An analysis of the 
distribution of variables for the respondents related to the size of 
the railroads was compared to the distribution of these variables in 
the entire population of railroads, and no important distributional 
differences were found. 

In addition to the data from the survey provided in this report, each 
survey question, along with responses to it, is presented in GAO-11- 
894SP, an electronic supplement to this report. 

Analysis of FRA Inspection and Enforcement Data: 

To determine the extent to which FRA conducts inspections of 
railroads' compliance with hours of service and hours of service 
recordkeeping requirements, we obtained information from FRA's 
Railroad Inspection System for PC. This system allows inspectors to 
enter inspection data via their personal computers in order to 
maintain electronic records. FRA provided data for all inspections 
conducted by FRA inspectors from fiscal year 2005 through fiscal year 
2010. We then excluded data for inspections of all entities that were 
not freight railroads. From the remaining data, we identified the 
number of hours of service and hours of service recordkeeping 
inspections that were conducted on freight railroads during this 6-
year period. We analyzed inspection results by class of railroad and 
determined the frequency with which deficiency findings identified 
during inspections resulted in an enforcement activity. 

To identify the enforcement actions FRA has taken in response to 
noncompliance with hours of service and hours of service recordkeeping 
requirements, we obtained data from FRA's Railroad Enforcement System. 
FRA's Office of Railroad Safety enters the information related to 
violations that have been recommended for citation against railroads 
and others in the Violation Generation Tracking System database, which 
populates the Railroad Enforcement System, which in turn is used by 
attorneys and staff to support the enforcement process. The data we 
obtained included all enforcement actions taken by FRA from the start 
of fiscal year 2005 through the end of fiscal year 2010. From this 
information, we identified all hours of service and hours of service 
recordkeeping violations involving freight railroads. We reviewed the 
data to identify the extent to which FRA pursues enforcement actions 
for hours of service violations, as well as the dollar amount it 
assesses in the form of fines and penalties. 

To assess the reliability of the inspection and enforcement data 
provided by FRA, we reviewed previous GAO reports about FRA's 
databases and FRA's efforts to ensure the data's reliability and 
conducted electronic testing of required data elements to identify 
omissions, anomalies, or obvious errors. In addition, we interviewed 
agency officials knowledgeable about the data quality control 
procedures and the data produced by the systems. We also determined 
whether the databases we used had been audited either internally or by 
external organizations. We determined that the data were sufficiently 
reliable for the purposes of this report. 

Additional Methodologies: 

To further address our objectives, we reviewed laws and regulations 
related to hours of service issues and reviewed various studies and 
other documents. To address impacts of RSIA's hours of service changes 
on the covered T&E workforce, including potential impacts on fatigue, 
we reviewed literature related to fatigue and work schedules and 
reviewed two reports prepared by FRA to validate the usability of the 
fatigue models FAST and FAID to assess the fatigue risk associated 
with railroad covered employee work schedules.[Footnote 66] These two 
reports provided information on such topics as how the models assess 
fatigue levels, assumptions used in making such assessments, how 
fatigue scores relate to the probability of accidents, and limitations 
of the model results. This information guided our use of the models to 
assess fatigue risk in the work schedules we reviewed. We also 
reviewed FRA's March 22, 2011, Notice of Proposed Rulemaking on new 
hours of service requirements for commuter and intercity passenger 
railroads, FRA's Regulatory Impact Analysis associated with this 
rulemaking, and the final hours of service rules which were issued in 
August 2011.[Footnote 67] In particular, we were interested in FRA's 
evaluation of fatigue risk associated with consecutive days worked and 
work performed during night hours. 

To address FRA's actions to ensure compliance with hours of service 
requirements, we reviewed documentation related to the National Rail 
Safety Action Plan, National Inspection Plan, and National Safety 
Program Plan. We also obtained selected FRA regional inspection plans 
to identify how inspection resources are allocated at the local level. 
Finally, we obtained data on petitions filed by railroads and others 
from May 2009 through June 2011 for waivers of compliance with hours 
of service requirements. These data included information on who filed 
petitions, when they were filed, and what their status was as of June 
2011. We verified these data with FRA and confirmed the status of each 
petition with FRA officials. 

To address our objectives, we also interviewed relevant individuals 
and organizations, including the following: 

* Federal officials, including those from the National Transportation 
Safety Board, FRA headquarters, and FRA regions 3, 4, 5, and 6. We 
selected these regional offices because we were already doing other 
work in the regions and the offices are geographically dispersed 
across the country. These four regional offices accounted for 60 
percent of the hours of service inspections conducted from fiscal year 
2005 through fiscal year 2010, and their territories cover all or 
parts of 23 states. We discussed with FRA the methods and procedures 
used to assess the fatigue risk in the railroad industry, the 
potential operational and administrative impacts of RSIA's hours of 
service changes on the railroad industry, and the processes and 
procedures FRA uses to ensure compliance with hours of service 
requirements. We also discussed FRA's actions to implement pilot 
projects related to hours of service and FRA's handling of petitions 
for waivers of compliance with hours of service requirements and the 
status of these petitions. 

* Fatigue and sleep research experts. We interviewed officials from 
the firms involved in developing the FAST and FAID models, the 
Institute of Behavioral Research and InterDynamics, Inc., 
respectively, as well as fatigue and sleep research experts. Our 
discussions with the model developers focused on how and why the 
models were developed, what assumptions were used in the modeling 
process, how we should use the models to assess fatigue risk in the 
railroad industry, and what limitations might be associated with the 
model results. After we acquired the models, officials from these 
companies also trained us in how to use the models and how to 
interpret their results. We also interviewed four experts in fatigue 
research. We spoke with these individuals about issues related to work 
and fatigue and factors relating to the potential for fatigue risk. We 
also solicited their views about fatigue models in general and the two 
fatigue models we acquired to analyze covered employee work schedules. 

* Railroad and railroad trade association officials. We interviewed 
officials from all 7 class I railroads, 6 class II railroads, and 6 
class III railroads as well as officials from a holding company that 
was the parent company for 39 class III railroads and 1 class II 
railroad. We discussed such issues as the effects of the hours of 
service changes on railroads and the covered workforce and the federal 
role in hours of service. We also discussed hours of service issues 
with officials from the Association of American Railroads, which 
represents the interests of class I railroads and the National 
Railroad Passenger Corporation, and the American Short Line and 
Regional Railroad Association, which primarily represents the 
interests of class II and III railroads. We also spoke with officials 
from the American Public Transportation Association about work they 
were doing to develop hours of service requirements for commuter and 
intercity passenger railroads. We were particularly interested in 
their views on the relationship between railroad work schedules and 
the potential for fatigue. 

* Representatives of labor organizations. We interviewed 
representatives from the Brotherhood of Locomotive Engineers and 
Trainmen, American Train Dispatchers Association, United 
Transportation Union, Transportation Communications Union, Brotherhood 
of Railroad Signalmen, National Conference of Firemen and Oilers, and 
the Transportation Trades Department of the AFL-CIO. These 
organizations represent various employees that would be covered by 
railroad hours of service requirements, including train and engine 
employees, signalmen, and dispatchers. According to these 
organizations, they represent over 100,000 employees covered by hours 
of service requirements. We solicited their views on the effects of 
RSIA's hours of service changes on their members, the benefits of 
these changes, and the federal role in monitoring and enforcing hours 
of service changes. We also solicited their views on waivers and 
exemptions to hours of service requirements for which railroads have 
applied. 

* We conducted this performance audit from April 2010 to September 
2011 in accordance with generally accepted government auditing 
standards. Those standards require that we plan and perform the audit 
to obtain sufficient, appropriate evidence to provide a reasonable 
basis for our findings and conclusions based on our audit objectives. 
We believe that the evidence obtained provides a reasonable basis for 
our findings and conclusions based on our audit objectives. 

[End of section] 

Appendix II: The Development of Biomathematical Models of Fatigue: 

This appendix discusses (1) why concerns about fatigue in the modern 
workplace have increased, (2) the nature of biomathematical models 
that have emerged to better understand sleep-work schedules and 
fatigue, and (3) GAO's use of biomathematical fatigue models for 
analyzing the effect on workers of RSIA's hours of service changes. 

Workplace Fatigue Is a Growing Concern in Modern Industrial Society: 

Over the past several decades, technology has enabled, and the 
globalization of society has increasingly come to expect, round-the- 
clock activities. Society has become "24/7." Planning for sleep is 
difficult when work schedules are unpredictable, and work that takes 
place outside normal business hours often requires people to sleep 
when humans are normally awake. These characteristics of the modern 
work world have led to a growing concern about human fatigue and its 
consequences in the workplace. These issues are particularly important 
to the rail industry, since rail workers often work on short notice 
and rail operations often occur at night. 

When a person does not get enough sleep, certain areas of the brain 
involved in cognition are affected, engendering fatigue and an 
associated state of diminished capacity. This diminished capacity can 
have a variety of ramifications that may be of concern. For example, 
when fatigued, humans have more difficulty maintaining attention, 
become less communicative, and have reduced situational awareness. 
They are then at greater risk of committing errors in their work, 
which can ultimately lead to more accidents. Concern about these 
effects has led to the development of tools for better understanding 
worker fatigue, predicting its extent, and mitigating its effects. 
[Footnote 68] 

Biomathematical Models Have Been Developed to Estimate Workplace 
Fatigue: 

Over the past several decades, a science has developed that examines 
the nature of human sleep and the effects of sleep deprivation. More 
recently--in about the past 20 years--a variety of researchers have 
developed tools that are designed to use data on individuals' sleep- 
wake patterns to estimate a variety of outcomes such as fatigue, 
cognition, and accident risk. Most of the current models are based on 
or informed by what is known as the "two-process model" of sleep 
regulation, developed in the early 1980s.[Footnote 69] 

Generally, the two-process model posits that alertness is a function 
of two primary factors:[Footnote 70] 

* The status of sleep/wake balance. The first factor rises and falls 
based on time spent sleeping and time spent awake. The model 
essentially posits that a person's alertness decays during waking 
hours and is restored with sleep and that the patterns of decay and 
restoration are reasonably predictable. The longer a person is awake, 
the more fatigued that person will become, and the associated 
reduction in alertness increases the risk of errors and accidents. 
Alertness can only be restored through sleep, and the model generally 
assumes that the first few hours of sleep contribute the most to 
recovery. That is, sleep intensity is greatest when sleep debt is at 
its greatest, which is during the first few hours of sleep. 

* Circadian influence. The second factor is related to circadian 
rhythm. Essentially, humans are hard-wired to sleep during the night 
and to be awake during the day. When people do sleep during the day, 
their rest is seldom as restorative as night sleep. First, it is 
apparently difficult to sleep during the day when core body 
temperatures are higher. Second, day sleep may be more prone to 
disruptions that limit its benefit. Third, the circadian pressure to 
sleep is highest at night, so humans tend to be less alert and more 
prone to lapses in attention at times when sleep normally occurs. 
Thus, fatigue--independent of the first factor, which addresses the 
extent of sleep deficit--tends to accrue more quickly when people work 
at night. 

While most of the current biomathematical models of fatigue 
incorporate these factors into their analysis, they differ in how the 
factors are structured. In particular, there is variation in the 
assumptions about function form and other mathematical underpinnings 
to the models. For example, there may be differences in how the rate 
of decay in cognition with waking hours is formulated, or the manner 
in which circadian factors are accounted for in the formulas. 
Moreover, the models vary in the specific outputs they provide. And 
finally, they vary in terms of the inputs necessary. Some require 
actual sleep histories, while others infer how much sleep a person 
would be likely to obtain from the person's work hours. 

The development of biomathematical fatigue models is very recent, and 
the models have critical limitations that are important for 
interpreting and using their outputs. The models provide a suggestion 
about the alertness of humans generally, not of individuals. 
Individuals vary widely in how they fatigue for a variety of reasons 
including differences in their personal circadian rhythms, their 
health, and social responsibilities. 

GAO Acquired Two Models for Use In Analyzing the Effects of Hours of 
Service Changes on Rail Worker Fatigue: 

GAO was asked to examine whether and how RSIA's changes to the hours 
of service laws affect rail worker fatigue. To do this, we determined 
that, despite the limitations of the current biomathematical fatigue 
models, there was merit in using them to study predicted fatigue based 
on railroad workers' history of work schedules before and after the 
rules were implemented. In particular, we determined, through 
discussions with several experts in sleep-fatigue research, that using 
the models to assess the change in scores for a set of workers after 
the new law was implemented was a reasonable use of these models 
because our focus is not on the scores of any particular workers, but 
rather on the trend in overall scores given changed scheduling 
patterns. 

Through discussion with Federal Railroad Administration (FRA) 
officials and others we determined that, at the time of our work, 
there were only two current biomathematical fatigue models that had 
been validated by FRA for use in assessing fatigue in the rail 
industry and that were appropriate to acquire for possible use in our 
analysis. We acquired both models. The first, the Fatigue Avoidance 
Scheduling ToolTM (FAST), was originally developed for military use by 
Dr. Steven R. Hursh, et al.; the second, the Fatigue Audit InterDyneTM 
(FAID) tool, was developed by Gregory D. Roach, Adam Fletcher, and 
Drew Dawson from the Centre for Sleep Research, University of South 
Australia. In particular, the FRA validation examined whether the 
models' predicted level of fatigue correlated with rail accidents 
deemed to have a "human" causal component, but not with rail accidents 
that had no identified human cause. FRA found this relationship for 
both models. 

Both of these models require similar data on railroad workers' work 
schedules and provide generally similar outputs. In particular, both 
models require data, by employee, on work shift start and end times 
for the period of time to be evaluated. The FAST model also requires 
information on the locations of each shift's start and end, and 
average commute times for those locations. Both models operate by 
taking this information and inferring the likely sleep employees are 
obtaining between their work shifts. After estimating how much time 
and at what time of day employees are awake and asleep, the models 
estimate--depending on the model--scores for elevated fatigue or 
reduced effectiveness. 

[End of section] 

Appendix III: GAO Contact and Staff Acknowledgments: 

GAO Contact: 

Susan A. Fleming, (202) 512-2834 or flemings@gao.gov: 

Staff Acknowledgments: 

In addition to the individual named above, Sara Vermillion, Assistant 
Director; Amy Abramowitz; Elizabeth Eisenstadt; Lorraine Ettaro; 
Patrick Fuchs; Kathleen Gilhooly; Christopher Jones; Richard 
Jorgenson; Mitchell Karpman; Amanda Miller; Dae Park; and Betsey Ward 
made key contributions to this report. 

[End of section] 

Footnotes: 

[1] A ton-mile is the transportation of 1 ton of freight 1 mile. 

[2] For purposes of this report, we define "shift" as a duty tour as 
described in 49 C.F.R. § 228.5: (1) The total of all periods of 
covered service and commingled service for a train employee or a 
signal employee occurring between two statutory off-duty periods 
(i.e., off-duty periods of a minimum of 8 or 10 hours); or (2) The 
total of all periods of covered service and commingled service for a 
dispatching service employee occurring in any 24-hour period. 

[3] Pub. L. No. 110-432, Div. A, title I, 122 Stat. 4848 (Oct. 16, 
2008). 

[4] Train employees include those engaged in or connected with the 
movement of trains and signal employees are those that install, 
maintain, or repair signal systems. Dispatching service employees 
(those that dispatch orders relating to train movement) are also 
covered by hours of service laws. See 49 U.S.C. § 21101. RSIA did not 
increase the statutory minimum off duty period for dispatching service 
employees. For purposes of this report we use the terms train 
employees and train and engine employees interchangeably. 

[5] 49 C.F.R §1.49(oo). 

[6] RSIA also maintained the prior hours of service requirements for 
the passenger rail industry for three years from the date of its 
enactment or until the effective date of new regulations covering the 
industry. The Secretary of Transportation was authorized to develop 
hours of service requirements for train employees engaged in the 
commuter and intercity passenger rail industry that differed from the 
freight railroad requirements. If these requirements were not in 
effect by October 2011, then RSIA provided that the freight railroad 
requirements would apply to these train employees at that time. FRA 
issued a Notice of Proposed Rulemaking on these requirements in March 
2011. 76 Fed. Reg. 16200 (Mar. 22, 2011). The final rule was issued in 
August 2011, to be effective October 15, 2011. 76 Fed. Reg. 50360 
(Aug. 12, 2011). 

[7] For economic regulatory purposes, the Surface Transportation Board 
(STB) divides the railroad industry into three classes based primarily 
on annual operating revenues. For 2009, this revenue threshold was at 
least $378.8 million for class I, at least $30.3 million for class II, 
and less than $30.3 million for class III. For accident and incident 
reporting, FRA divides the railroad industry into three groups based 
on the total work hours reported annually. Group 1 railroads are the 
same as class I railroads. Group 2 railroads report 400,000 total work 
hours or more annually but are not class I railroads, and group 3 
railroads report less than 400,000 total work hours. According to FRA 
officials, groups 2 and 3 are roughly equivalent to class II and III 
railroads. In this report, we generally refer to railroads by class 
rather than by group--even though our selection of railroads for work 
schedule and survey purposes was based on FRA's groups, not classes--
since class is a more common identifier. The six class II railroads 
were Belt Railway of Chicago, Consolidated Rail Corporation, Indiana 
Harbor Belt Railroad, Paducah & Louisville Railroad, Port Terminal 
Railroad, and Wheeling & Lake Erie Railroad. 

[8] This is the same as a p-value < 0.05. 

[9] A train-mile is the movement of a train a distance of 1 mile. 

[10] A train accident is any collision, derailment, fire, explosion, 
act of God, or other event involving operation of railroad on-track 
equipment (standing or moving) that results in damages greater than 
the current reporting threshold to railroad on-track equipment, 
signals, track, track structures, and roadbed. The reporting threshold 
for calendar year 2011 is $9,400. 

[11] Switching is a railway service that is performed under yard rules 
and regulations and involves, among other things, changing the 
position of railcars for purposes of loading, unloading, or weighing. 

[12] In this report, we do not attempt to draw any correlations 
between safety outcomes (such as changes in accident rates) and 
changes to hours of service requirements contained in RSIA. 

[13] The third leading cause of train accidents was a category titled 
"miscellaneous causes." Miscellaneous causes can include such things 
as environmental conditions (e.g., snow, ice, and dense fog), loading 
procedures, and highway-rail grade crossing accidents. According to 
FRA, crossing accidents are analyzed as a separate category and 
excluded from FRA analyses of train accidents, and usually involve 
motorist error. 

[14] These changes included revisions to 49 C.F.R. § 217.9 and the 
addition of 49 C.F.R. Part 218, Subpart F. 

[15] Pub. L. No. 59-274, 34 Stat. 1415 (Mar. 4, 1907). The original 
law covered both train employees and dispatching service employees, 
with respect to common carriers by railroad engaged in interstate or 
foreign commerce. Although RSIA establishes the maximum number of on-
duty hours and minimum rest periods for covered employees, it 
authorized the Secretary of Transportation to issue regulations to 
reduce the maximum number of hours an employee may be required or 
allowed to go or remain on duty or to increase the minimum number of 
hours an employee may be required or allowed to rest, as well as other 
regulations to improve safety and reduce employee fatigue. 

[16] Pub. L. No. 91-169, 83 Stat. 463 (Dec. 26, 1969). 

[17] Pub. L. No. 94-348 § 4, 90 Stat. 817, 819 (July 8, 1976). RSIA 
further extended the coverage of hours of service requirements to 
certain contract and subcontract signal employees. Pub. L. No. 110-
432, Div. A., § 108(a). 

[18] 49 U.S.C. § 21103(a)(4). 

[19] The request for the approval of a pilot program also includes a 
request for a wavier of compliance with hours of service requirements 
49 U.S.C. § 21108(b). 

[20] Positive train control is a communications-based train control 
system designed to prevent certain types of train accidents. For more 
information about this topic, see GAO, Rail Safety: Federal Railroad 
Administration Should Report on Risks to the Successful Implementation 
of Mandated Safety Technology, GAO-11-133 (Washington, D.C.: Dec. 15, 
2010). 

[21] We previously reported on FRA's safety oversight approach in 
January 2007. See Rail Safety: The Federal Railroad Administration Is 
Taking Steps to Better Target Its Oversight, but Assessment of Results 
Is Needed to Determine Impact, GAO-07-149 (Washington, D.C.: Jan. 26, 
2007). 

[22] FRA launched the National Rail Safety Action plan in May 2005 
because, although the annual number of train accidents had been 
declining since 1995, the rate of accidents had not shown substantive 
improvement and serious train accidents were continuing to occur. 

[23] Other key agency plans focused efforts on the highest priority 
risks related to train accidents including the department's rulemaking 
agenda, strategic plan and annual performance plan, and FRA's 
performance budget. 

[24] FRA inspectors conduct two types of inspections related to hours 
of service--an hours of service inspection to determine if covered 
employees' work hours fall within RSIA's maximum time on duty and 
minimum time off duty limits and an hours of service recordkeeping 
inspection to assess railroads' compliance with applicable FRA 
recordkeeping regulations. 

[25] Civil penalties are FRA's primary enforcement tool, though the 
agency may also seek to impose criminal penalties if warranted. 

[26] 49 U.S.C. § 21303(a)(2). 

[27] 49 U.S.C. §§ 21102(b), 21103(a)(4). 

[28] The Rail Safety Act of 1970 (Pub. L. No. 91-458, title II, 84 
Stat. 971 (Oct. 16, 1970)) authorized the states to work with the FRA 
to enforce federal railroad safety regulations. Currently, the program 
includes inspectors from 30 states. 

[29] Dollar amounts have not been adjusted for inflation. 

[30] A revenue ton-mile is 1 ton of revenue freight transported 1 mile. 

[31] RSIA permits railroads to petition for a waiver of compliance to 
hours of service requirements. Further discussion of petitions for 
waivers of compliance can be found later in this report. 

[32] Hours available for rest was determined by calculating the total 
hours in the month of May and subtracting the total hours worked, 
which includes both covered and noncovered work performed by class I 
and class II T&E employees contained in the May 2008 and May 2010 
railroad provided data. 

[33] As noted above, unless otherwise specified, all statistics (or 
coefficient estimates) mentioned throughout this report are 
significantly different from zero with probability of error less than 
0.05 (or p-value < 0.05). 

[34] For the purposes of our analysis, we define high risk of fatigue 
as a FAID score of 70 or higher. See appendix I for further discussion 
on the fatigue score decision. 

[35] We use the term fatigue model to refer to the FRA-validated 
biomathematical model we used for our primary analysis. To analyze 
fatigue risk, these models incorporate a number of work-related 
factors, such as shift length, time of day or night when the shift 
occurs, and number of consecutive days at work. The models do not 
account for non-work-related factors like chronic illness or sleep 
disorders. 

[36] All fatigue modeling discussed in this section was conducted 
using the FAID model. 

[37] FRA, Hours of Service of Railroad Employees; Substantive 
Regulations for Train Employees Providing Commuter and Intercity Rail 
Passenger Transportation; Conforming Amendments to Recordkeeping 
Requirements Notice of Proposed Rulemaking, Regulatory Impact 
Analysis, p. 23. Docket No. FRA-2009-0043 Notice No. 1, RIN 2130AC15 
(Washington, D.C.: Mar. 10, 2011). 

[38] For the purpose of our analysis, we defined night work as work 
performed between the hours of 11 p.m. and 6 a.m. 

[39] Simon Folkard and Torbjorn Åkerstedt, Trends in the Risk of 
Accidents and Injuries and Their Implications for Models of Fatigue 
and Performance, Aviation, Space and Environmental Medicine, March 
2004, Vol. 75, No.3, Section II, p. A163 and p. A165. 

[40] We did not analyze work schedules for class III railroads 
because, as noted, most were paper-based. As a result, we were unable 
to conduct a similar analysis of night work and fatigue risk for class 
III covered T&E employees. 

[41] 76 Fed. Reg. No. 156, p. 50364 (Aug. 12, 2011). 

[42] Specifically we found the correlation coefficient of 0.53 between 
hours worked at night and the incidence of workers spending at least 
20 percent of their hours worked at a high fatigue level. 

[43] Not all respondents to the survey answered each of the questions. 
An analysis of the distribution of variables for the respondents 
related to the size of the railroads was compared to distribution of 
these variables in the entire population of railroads and no 
significant distributional differences were found. 

[44] Officials we spoke with from a class II railroad said that before 
RSIA took effect, railroads could have crews back on duty in 8 to 10 
hours and could call employees within 2 hours of their on-duty time, 
as set forth in collective bargaining agreements. After RSIA, with the 
minimum rest requirement increased from 8 hours to 10 hours and the 
new prohibition against calling employees during their rest time, 
crews cannot be back on duty before 12 hours or more. According to 
FRA, it could take 12 hours or more for an employee to return to work 
if they are not given a time to report for work again at the end of 
their prior shift and are given 2 hours to report for work after 
receiving a duty call, as is common under many collective bargaining 
agreements. 

[45] For a train employee, covered service is the portion of the 
employee's time on duty during which the employee is engaged in, or 
connected with, the movement of a train. Noncovered service is 
generally time spent performing tasks for a railroad that are not 
covered by hours of service requirements, such as managing the 
business. 

[46] In responding to our survey, all 7 class I, 9 of 14 class II, and 
54 of 151 class III (about 36 percent) railroads reported taking 
employees out of service as a direct result of RSIA's requirements. 

[47] Officials we spoke with at one railroad said its threshold varied 
by crew pool, and officials at another railroad said it did not have a 
set threshold but rather its crew calling and scheduling system 
prevented an employee from starting a tour of duty when the average 
trip time might cause the employee to reach or exceed the total 
monthly work hour cap. 

[48] This information is similar to the results of a study FRA 
reported in November 2009. Designed to characterize the work/rest 
schedules and sleep patterns of U.S. railroad T&E personnel, this 
study used a background survey and daily log over a 14-day period for 
a random sample of 809 T&E workers and then extrapolated the daily log 
data to a 30-day period. According to FRA, 12 of 232 respondents (5.2 
percent) whose daily logs had complete work histories would have 
likely exceeded RSIA's monthly cap had it been in place at the time of 
the FRA study. See Work Schedules and Sleep Patterns of Railroad Train 
and Engine Service Workers, U.S. Department of Transportation, Federal 
Railroad Administration, DOT/FRA/ORD-09/22 (November 2009). 

[49] Yardmasters coordinate the activities of workers engaged in 
railroad yard operations. These activities include making up or 
breaking up trains and switching inbound or outbound traffic to 
specific sections of a rail line. According to FRA, depending on the 
exact duties performed, yardmasters may be train employees, 
dispatching service employees, both, or neither. 

[50] The National Industrial Transportation League reported it has 
over 600 company members, about a third of which use rail to ship or 
receive goods. The association surveyed these rail users and received 
responses from 28 of them. 

[51] As identified earlier, switching is a railway service that is 
performed under yard rules and regulations and involves, among other 
things, changing the position of railcars for purposes of loading, 
unloading, or weighing. "Timing out on the clock" occurs when a 
railroad crew works up to the maximum time allowed by the hours of 
service requirements (currently 12 hours for T&E crews). 

[52] In addition, 1 class III railroad responded that the number of 
claims decreased and 1 class III railroad responded "don't know" to 
this question. 

[53] As previously indicated, each FRA inspector belongs to one of 
five expertise areas, called disciplines--operating practices, track, 
hazardous materials, signal and train control, and motive power and 
equipment (such as locomotives and tank cars), and according to FRA, 
each discipline corresponds to certain rail safety or hazardous 
materials regulations, orders or statutes. 

[54] Specifically, FRA obtains and analyzes the work schedules of all 
employees involved in an incident or accident for 11 days prior to the 
accident or incident. FRA analyzes these work schedules using a 
fatigue model to determine if fatigue may have been a contributing 
factor to the accident or incident. Such a review of an employee's 
work schedule may also reveal noncompliance with the hours of service 
laws. 

[55] Pub. L. No. 110-432, Div. A., § 108 (e)(1). RSIA requires the 
Secretary of Transportation to report to the Senate Committee on 
Commerce, Science and Transportation and the House Committee on 
Transportation and Infrastructure no later than December 31, 2012, or 
if no projects are approved prior to that, no later than 6 months 
after the completion of a pilot project. Pub. L. No. 110-432, Div. A., 
§ 110. 

[56] For railroads without labor organizations, FRA requires that the 
railroad consult with directly affected covered service employees. 

[57] In both of these cases, the railroads in question will split 
their workforces when calculating their covered employees' 276-hour 
monthly work hour cap. Half of the workforce will have its hours 
calculated from the first to the last of the month, and the other half 
will have its hours calculated from the 15th of the month to the 14th 
of the following month. The intent of this effort is to avoid the 
possibility that a large portion of the railroad's covered employees 
will be at or near the cap at the end of the month, and be unavailable 
to work. This assurance will allow railroads to continue their 
operations uninterrupted. 

[58] 49 U.S.C. § 21103(a)(4). 

[59] In addition to these 17 petitions, FRA approved one railroad's 
petition for an emergency waiver of the hours of service requirements 
in order to support flood relief efforts along the Missouri River in 
June 2011. Because this was an emergency waiver, we excluded it from 
our analysis. The ASLRRA petition originally covered 301 railroads, 
but some did not meet the joint filing requirements to obtain the 
waiver. FRA determined that 142 railroads met the requirements. 

[60] These railroads were the Belt Railway of Chicago, Consolidated 
Rail Corporation, Indiana Harbor Belt Railroad, Paducah & Louisville 
Railroad, Port Terminal Railroad, and Wheeling and Lake Erie Railroad. 

[61] For the purposes of our analysis, we defined night work as work 
preformed between the hours of 11 p.m. and 6 a.m. 

[62] FRA, Procedures for Validation and Calibration of Human Fatigue 
Models: The Fatigue Audit InterDyne Tool, DOT/FRA/ORD-10/14 
(Washington, D.C.: November 2010). 

[63] Accidents and incidents is a term used to describe an entire list 
of reportable events, including fatalities, injuries, and illnesses; 
collisions, derailments, and similar accidents involving the operation 
of on-track equipment causing reportable damage above an established 
threshold; and impacts between railroad on-track equipment and highway 
users at crossings. The 2011 damage threshold for reporting accidents 
and incidents is $9,400. 

[64] The monthly report must be submitted within 30 days following the 
month to which it applies. See 49 C.F.R. §§ 225.11 and 225.19. See 49 
C.F.R. Part 225 for accident and incident reporting requirements. 

[65] FRA also exempts certain railroads that haul passengers inside an 
installation that is insular--that is, generally has no public highway-
rail grade crossings or bridges over public roads or waters used for 
commercial navigation and does not share a common corridor with 
another railroad. 

[66] FRA, Validation and Calibration of a Fatigue Assessment Tool for 
Railroad Work Schedules, Final Report DOT/FRA/ORD-08/04 (Washington, 
D.C.: November 2008), and FRA, DOT/FRA/ORD--10/14 (Washington, D.C.: 
November 2010). 

[67] 76 Fed. Reg. 50630 (Aug. 12, 2011). 

[68] David Neri, "Preface: Fatigue and Performance Modelling Workshop, 
June 13-14, 2002," Aviation, Space, and Environmental Medicine, Vol. 
75, No. 3, Section II (March 2004). 

[69] Drew Dawson, Y. Ian Noy, Mikko Härmä, Torbjorn Åkerstedt, and 
Gregory Belenky, "Modelling Fatigue and the Use of Fatigue Models in 
Work Setting," Accident Analysis and Prevention 43 (2011) 549-564, and 
Melissa M. Mallis, Sig Mejdal, Tammy T. Nguyen, and David F. Dinges, 
"Summary of the Key Features of Seven Biomathematical Models of Human 
Fatigue and Performance," Aviation, Space, and Environmental Medicine, 
Vol. 75, No. 3, Section II (March 2004). 

[70] A third factor, referred to as "sleep-inertia" is also included 
in some so-called "three-process models of sleep." This factor adjusts 
for a lag in fully obtaining predicted alertness just after awakening. 

[End of section] 

GAO's Mission: 

The Government Accountability Office, the audit, evaluation and 
investigative arm of Congress, exists to support Congress in meeting 
its constitutional responsibilities and to help improve the performance 
and accountability of the federal government for the American people. 
GAO examines the use of public funds; evaluates federal programs and 
policies; and provides analyses, recommendations, and other assistance 
to help Congress make informed oversight, policy, and funding 
decisions. GAO's commitment to good government is reflected in its core 
values of accountability, integrity, and reliability. 

Obtaining Copies of GAO Reports and Testimony: 

The fastest and easiest way to obtain copies of GAO documents at no 
cost is through GAO's Web site [hyperlink, http://www.gao.gov]. Each 
weekday, GAO posts newly released reports, testimony, and 
correspondence on its Web site. To have GAO e-mail you a list of newly 
posted products every afternoon, go to [hyperlink, http://www.gao.gov] 
and select "E-mail Updates." 

Order by Phone: 

The price of each GAO publication reflects GAO’s actual cost of
production and distribution and depends on the number of pages in the
publication and whether the publication is printed in color or black and
white. Pricing and ordering information is posted on GAO’s Web site, 
[hyperlink, http://www.gao.gov/ordering.htm]. 

Place orders by calling (202) 512-6000, toll free (866) 801-7077, or
TDD (202) 512-2537. 

Orders may be paid for using American Express, Discover Card,
MasterCard, Visa, check, or money order. Call for additional 
information. 

To Report Fraud, Waste, and Abuse in Federal Programs: 

Contact: 

Web site: [hyperlink, http://www.gao.gov/fraudnet/fraudnet.htm]: 
E-mail: fraudnet@gao.gov: 
Automated answering system: (800) 424-5454 or (202) 512-7470: 

Congressional Relations: 

Ralph Dawn, Managing Director, dawnr@gao.gov: 
(202) 512-4400: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7125: 
Washington, D.C. 20548: 

Public Affairs: 

Chuck Young, Managing Director, youngc1@gao.gov: 
(202) 512-4800: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7149: 
Washington, D.C. 20548: