Body
A new analysis by the Howard Center for Investigative Journalism at the University of Maryland says rail safety reforms repeatedly recommended after major crashes have often stalled inside the U.S. regulatory process. The center said human errors and track defects were factors in more than 3,000 rail accidents over the last decade, killing 23 people and injuring nearly 1,200, while the Federal Railroad Administration did not implement most of the safety recommendations that flowed from federal crash investigations. The center’s findings describe a pattern in which industry opposition and regulatory inaction combine to delay changes meant to prevent future derailments and collisions.
The Howard Center said it examined recommendations issued by the National Transportation Safety Board from 2015 through 2024 and found the Federal Railroad Administration had only fully implemented five of 81 recommendations. The center said its review points to an industry that, in its description, uses money and influence to stymie federal safety reforms. It said the rail industry is also lobbying President Donald Trump’s administration to reduce track inspection and repair requirements and loosen rules intended to prevent crew fatigue.
A Federal Railroad Administration spokesperson, Warren Flatau, said in late September that the agency had 102 open safety board recommendations, some dating back as far as 1998, and “is currently taking action or planning to take action” on more than 70% of them. Flatau said the agency planned no immediate action on the rest, generally because officials determined recommendations were infeasible or could be addressed with existing regulations. The analysis said the agency did not respond to requests for further information about what actions it was taking.
One example cited by the Howard Center involved a 2021 derailment of Amtrak’s Empire Builder passenger train in Joplin, Montana, in which three people were killed as the train derailed just before 4 p.m. on Sept. 25. The center said the NTSB investigation blamed the tragedy in part on worn-down rail and other track defects that investigators said a BNSF Railway inspector likely missed because of workload. The analysis also said the NTSB pointed to the Federal Railroad Administration’s failure to set rules for track replacement and workloads, measures the NTSB said the agency had recommended repeatedly after previous accidents.
The center traced the same issues to older crashes, including a 1980 derailment in Muldraugh, Kentucky, involving a train carrying vinyl chloride. In that case, the NTSB found the accident was caused in part by dangerously worn-down rail, and it said regulators did not address proposed limits on how much rail wear is allowed before replacement. The analysis said the NTSB’s earlier call for rail wear limits would have required BNSF to replace the rail involved in the Montana derailment, and it described how an inspector documented concerns but did not walk the tracks in the period before the crash.
In the period leading up to the Joplin derailment, the Howard Center said investigators found no reported walking inspection in nearly a year and that an inspector missed defects after riding over the area rather than stopping to examine it more closely. The analysis said investigators also found that, on the day of the inspection, the inspector had worked 16 hours inspecting 127 miles of track and that the NTSB faulted BNSF’s safety culture for not adequately managing employee workloads. The center said the NTSB has warned for years that fatigued workers can make mistakes.
The Howard Center also said fatigue risk management reforms faced similar delays. It cited a fatal 2004 Union Pacific collision in Macdona, Texas, where the Howard Center said the NTSB found an engineer was operating with less than two hours’ sleep and blew through a signal light, contributing to a crash that left multiple people dead and others in severe respiratory distress. The center said the NTSB concluded the engineer’s fatigue was triggered by chronic lack of sleep, a disrupted sleep cycle and long work shifts, and it said federal regulators had broad discretion over crew scheduling.
The analysis said federal action on fatigue-related rules took years to complete. It said Congress, via the Rail Safety Improvement Act, mandated implementation of fatigue risk management plans in freight rail and required railroads to screen employees for sleep disorders and provide fatigue training. The Howard Center said it took more than 15 years for the Federal Railroad Administration to finish approving fatigue plans for all railroads, and it said its own analysis of NTSB investigations found fatigue-related accidents killed at least nine people and injured more than 300 from 2015 through 2024.
In addition to track inspection requirements and fatigue protections, the Howard Center highlighted industry efforts to slow or narrow regulatory oversight. It said the rail industry’s largest trade group, the Association of American Railroads, pushed to reduce human inspection requirements in favor of increased automation. The analysis said that in May the AAR suggested reducing the frequency of required human track inspections and argued that automation would “improve safety and reduce regulatory costs,” and that on Dec. 5 federal regulators approved a waiver allowing participating railroads to cut weekly visual inspections in half with new safety and reporting requirements.
The Howard Center said the NTSB considers automation a supplement rather than a replacement for human inspection because neither method is 100% effective. It said BNSF told investigators that automation allows the company to detect and repair defects more quickly, while a rail inspector for Norfolk Southern, Lance Marston, said technology excels at identifying certain defects not always visible to the human eye, such as rails being too far apart. Marston said, “There’s just no replacement for getting out there and putting boots on the ground,” and the Howard Center said Christie Lee, a BNSF track inspector, described higher workloads after automation reduced staffing and extended territories.
The Howard Center further described the influence of the rail industry in regulatory timelines, saying the sector delayed adoption of positive train control for decades and pushed Congress to postpone deadlines after early 2000s tragedies. It said positive train control ultimately became fully compliant in 2020, and that by then an internal NTSB tally showed preventable accidents had claimed 300 lives and injured 6,800 people. The center said an AAR statement attributed the delay to the “complexity and scale” of technological development.
As the Trump administration invited public input on how to reduce federal bureaucracy earlier this year, the analysis said the AAR responded with a spreadsheet itemizing 80 regulations for modification or repeal. It said the deregulatory agenda would reduce oversight of an industry that carries billions of passengers and transports millions of gallons of toxic chemicals through thousands of communities. The Howard Center said items on the AAR’s list included repealing a requirement for minimum two-person crews and reducing administrative requirements for fatigue management plans, and it said the AAR described its aim as cutting through regulatory red tape in a “manner that is consistent with railroad safety.”
All three of the major American freight railroads included in the Howard Center’s story—BNSF, Norfolk Southern and Union Pacific—sent statements saying they are committed to safety, the analysis said. The Howard Center report, produced with data and reporting contributions from University of Maryland staff including Mary Burke, Taylor Nichols, Adriana Navarro and April Quevedo, framed its findings as a recurring cycle of industry opposition, regulator inaction, congressional capitulation and tragedy.
At the center of the report is a question of follow-through on recommendations issued after investigations by the National Transportation Safety Board—recommendations that, the analysis says, have too often remained unimplemented long enough for serious crashes to recur.