Rail failure statistics offer more than a count of trains that leave the tracks. They provide a way to understand where railroad systems experience failures, which conditions appear most often, and how those patterns change over time.
In 2026, the latest Federal Railroad Administration (FRA) records provide another opportunity to examine those patterns, although the currently published figures cover only part of the year.
For railroad operators, maintenance teams, engineers and safety professionals, the value of rail failure statistics comes from connecting several types of information rather than looking at derailments in isolation. Accident records show what happened. Inspection records reveal conditions identified before or between accidents. Operational data provide the amount of railroad activity against which those failures should be measured.
That distinction becomes particularly important when current FRA data are compared with the longer historical record.
What Counts as a Railroad Failure in FRA Data?
The first step in interpreting rail failure statistics is understanding what the FRA actually records.
FRA's train accident database is built largely around Form FRA 6180.54, Rail Equipment Accident/Incident. Reportable events involving railroad on-track equipment, including derailments, collisions, fires and explosions, are submitted through this reporting system.
The FRA safety data system allows accident information to be examined by factors such as accident type, cause, track type, track class, railroad and location.
An inspection defect represents something different. A defect is a condition identified during an inspection that does not meet an applicable requirement or standard. It does not mean that a derailment has occurred, and not every defect represents the same level of risk.
This creates an important distinction:
| FRA Measure | What It Represents |
|---|---|
| Derailment | An accident outcome |
| Track or equipment defect | A condition identified during inspection |
| Train-miles | Exposure to railroad activity |
| Injuries, fatalities and damage | Consequences of an accident |
The distinction matters because rail failure statistics can become misleading when accident outcomes and inspection findings are treated as though they were the same measurement.
A derailment is a visible failure. A defect can be an earlier indication of a condition that requires attention.
The Long-Term Derailment Picture Is More Useful Than a Single-Year Number
Recent FRA records show considerable year-to-year movement. According to historical FRA data summarized by the Congressional Research Service, derailments excluding highway-rail crossing incidents totaled:
- 2019: 1,344
- 2020: 1,117
- 2021: 1,118
- 2022: 1,229
- 2023: 1,310
- 2024: 1,113
- 2025: 976
These figures provide useful context, but they should not be interpreted as a simple safety scorecard. A reduction in derailments does not automatically mean that underlying risk has fallen by the same percentage.
The operating environment changes from year to year. Train volumes, train lengths, traffic density, infrastructure condition, operating practices and inspection activity can all affect the number and nature of reported events.
The 2026 figure requires an additional qualification. FRA's public safety data currently contain information through June 30, 2026. FRA updates its data monthly, with railroads generally required to submit reportable information within 30 days after the end of the relevant month and FRA given additional time to process and publish it.
As a result, 2026 should be treated as a year-to-date observation, not a completed annual result.
A fair current-year comparison therefore compares January-to-June 2026 with the corresponding January-to-June period of 2025 rather than comparing a partial year with a full year.
This leads to a more important question: does the number of derailments actually reflect the amount of railroad activity taking place?
Why Railroad Activity Matters When Interpreting Failure Statistics
Raw rail failure statistics can hide an important part of the safety picture: exposure.
A railroad operating thousands of additional train-miles has more opportunities for an accident than a railroad operating fewer trains.
For that reason, researchers have long used exposure-normalized measures rather than relying solely on accident counts.
A basic measure can be expressed as:
Derailment rate = Number of derailments ÷ Train-miles × 1,000,000
The resulting figure represents derailments per million train-miles.
This approach makes comparisons between years and operating environments more meaningful. It also reflects the methodology used in established railroad safety research.
Liu, Saat and Barkan analysed FRA derailment records from 2001 through 2010 by cause, track type, derailment speed and number of cars derailed. Their work treated accident frequency as part of a broader quantitative risk-analysis process rather than simply counting events.
More recent research continues this approach. A 2026 study by C. Tyler Dick and Peter F. Swan examined normalized mainline freight derailment rates for 2018–2023 and compared them with 2012–2017. The research considered both frequency and severity while examining changes associated with railroad operating practices.
Meaningful rail failure statistics should therefore answer two separate questions:
- How many failures occurred?
- How many failures occurred relative to the amount of railroad activity?
The second question often reveals a clearer trend than the raw accident count alone.
What Is the Number One Cause of Train Derailments?
There is no single cause that explains every derailment across every railroad environment. However, FRA data and historical research consistently identify several major cause groups.
These include:
- Track-related causes: rail or track-condition problems.
- Mechanical causes: wheel, bearing, axle and other equipment failures.
- Human-factor causes: operating, switching and train-handling errors.
- Signal-related causes: signalling or train-control conditions.
- Other causes: circumstances that do not fit neatly into the major categories.
Historical research provides an important benchmark. In the 2001–2010 FRA dataset analysed by Liu, Saat and Barkan, broken rails or welds represented the leading derailment cause across main, yard and siding track categories.
At speeds below 10 mph, track and human-factor causes such as improper train handling, braking operations and improper switch use were prominent. At speeds above 25 mph, equipment-related causes such as bearing failures, broken wheels and axle or journal defects became more important.
This highlights an important point about rail failure statistics:
The most frequent cause is not necessarily the most consequential cause.
A low-speed switching derailment and a high-speed mainline derailment may each appear as a single accident in a database, yet their potential consequences can be dramatically different.
Severity therefore belongs beside frequency.
The number of cars derailed, injuries, fatalities, property damage and hazardous-material involvement can all change the significance of an event even when the accident count remains unchanged.
This is also where the limitations of accident-only analysis become clear. A derailment database describes failures that have already occurred. It tells us much less about conditions that existed before the event.
That is where inspection data become important.
From Derailments to Defects: What Happens Before a Failure?
The second side of rail failure statistics comes from inspection activity.
FRA's FY2025 Annual Enforcement Report recorded 269,141 defects across railroads and other inspected entities, including 249,423 railroad-only defects. FRA also reported 2,734,278 inspected units across all entities and 2,649,178 railroad-only units.
These numbers should not be interpreted as 269,141 imminent derailments.
Inspection defects cover a wide range of conditions. Some may be relatively minor, while others can identify conditions requiring prompt corrective action. The number of recorded defects also depends on how much equipment and infrastructure was inspected and how extensively inspections were conducted.
For that reason, the rate and type of defects are generally more informative than a raw national total.
The distinction between accident outcomes and inspection findings raises a more useful question: can inspection conditions provide an earlier indication of where failure risk may be developing?
Which Defects Appear Most Often in FRA Inspection Records?
Inspection records add another dimension to rail failure statistics because they capture conditions before they necessarily become accidents.
FRA enforcement information separates inspection findings into areas including track, motive power and equipment, safety appliances, roadway worker protection, signal-related requirements and other regulatory categories.
In FY2025, FRA recorded 2,927 recommended violation defects in motive power and equipment categories, including safety-appliance, freight-car, locomotive and passenger-equipment standards. It also recorded 522 recommended violation defects associated with track and related standards.
These figures require context. A railroad with more inspections can naturally produce more recorded defects. Similarly, a large railroad with more locomotives, freight cars and track mileage has more opportunities for inspectors to identify conditions requiring correction.
Useful rail failure statistics should therefore consider measures such as:
- Defects per inspection
- Defects per inspected unit
- Defects per inspection day
- Defects relative to network or equipment exposure
This avoids turning the largest raw number into an automatic ranking of the least safe railroad.
The more useful question is whether particular defect patterns repeatedly appear alongside particular types of failures.
Can Inspection Defects Provide an Early Warning of Derailments?
This is where derailment and defect data become especially valuable when viewed together.
Conceptually, the relationship can be represented as:
Inspection finding → unsafe or deteriorating condition → failure mechanism → derailment
However, this sequence should not be treated as proof of causation. An inspection defect does not automatically cause a later derailment, and a statistical relationship between two annual measures does not establish that one directly caused the other.
A stronger analysis would compare defect and accident information at several levels.
For example, annual track-defect rates could be compared with annual track-caused derailment rates. Equipment-defect rates could be compared with equipment-related derailments. A time-lagged analysis could also compare defects recorded during one year with derailments occurring during the following year.
This type of analysis moves rail failure statistics beyond simple accident counting and toward the concept of leading safety indicators.
The approach is consistent with the direction of modern railroad risk research. A 2026 study examining FRA Rail Equipment Accident data found that the database has supported decades of safety and risk research while also identifying limitations in data resolution.
The study recommended improvements such as more detailed information on track geometry, train consist, railcar loading status and connections to external databases containing information such as weather and hazardous-material conditions.
This limitation matters because broad datasets can reveal relationships, but detailed prediction requires detailed information about time, location, equipment condition, track condition and operating circumstances.
How Does FRA Data Help Analyse Railroad Failures?
The FRA Safety Data Portal is the primary public source for current U.S. railroad safety information. It provides train accident reports, casualty information, operational data, downloadable datasets and other safety reports.
The train-accident section provides information from Form 6180.54 and allows analysis by accident type, accident cause, track type, track class, railroad, state, county and other characteristics.
FRA also provides operational information such as train-miles, employee hours, yard-switching miles, passengers transported and passenger-miles.
One data-handling issue deserves particular attention. Multiple railroads can submit Form 54 reports for the same physical accident when several railroads are involved.
Counting every raw report as a separate accident can therefore overstate the number of events. FRA provides a unique-train-accident dataset and summary reports that select one report to represent each accident.
For researchers and industry analysts, that distinction can make the difference between a reliable trend and a misleading one.
What Do the 2026 Rail Failure Statistics Mean for Railroad Safety?
The current rail failure statistics point toward a broader way of evaluating railroad safety.
The first measure is still the derailment count. It provides a straightforward picture of how many reportable events occurred.
The second is the derailment rate, which places those accidents against railroad activity.
The third is the defect picture, which shows what inspectors are finding across infrastructure and equipment.
Together, these measures answer different questions:
- Accident data show what happened.
- Inspection data show what conditions were found.
- Operational data show how much activity was exposed.
The combination provides a more useful safety picture than any one measurement alone.
Final Thoughts
The value of rail failure statistics extends well beyond counting how many trains have derailed. Strong analysis connects derailment frequency with operational exposure, examines the causes behind those events, separates different operating environments and considers inspection conditions identified before accidents occur.
The latest FRA information provides a useful 2026 snapshot, but the currently published data represent only part of the calendar year. That makes year-to-date comparisons more appropriate than attempts to declare a completed full-year trend.
The larger opportunity lies in combining datasets. Derailment records show the final outcome. Inspection records reveal conditions that may require attention. Operational data provide the context needed to measure exposure.
Together, these sources can support a more practical understanding of where railroad failure risks develop and how safety programmes can respond.
As inspection technologies, condition monitoring and data analytics continue to improve, rail failure statistics can move beyond describing past accidents and become increasingly useful for identifying patterns that deserve attention before the next failure occurs.
Key Takeaways
- FRA derailment records provide an important but incomplete picture of railroad safety performance.
- Annual accident totals become more meaningful when compared with actual railroad operating exposure.
- Mainline and yard derailments represent different operating environments and can have different safety consequences.
- Track conditions remain an important component of the railroad derailment cause picture.
- Equipment failures and human factors also contribute significantly to railroad accident patterns.
- FRA inspection defects represent identified conditions rather than completed accident events.
- Defect rates generally provide more useful comparisons than large absolute defect totals.
- FRA Form 54 records require careful handling when individual accidents are being counted.
- Published 2026 FRA figures represent year-to-date conditions rather than a completed calendar year.
- Combining accident, inspection, severity and operational data provides a broader railroad safety perspective.
Frequently Asked Questions
What was the worst train accident in U.S. history?
The Great Train Wreck of 1918 is widely regarded as the deadliest train accident in U.S. history. The collision occurred near Nashville, Tennessee, when two passenger trains collided head-on, killing 101 people and injuring more than 170.
The disaster contributed to greater attention on railroad operating practices, signalling and train-control safety.
Which is the safest train in the world?
There is no single train officially recognised as the “safest train in the world.” Rail safety depends on infrastructure, signalling, maintenance, operating practices, equipment condition and regulatory oversight.
Modern high-speed and passenger rail systems can achieve very strong safety records, but comparing individual trains without considering their operating environment can produce a misleading result.
What causes most train deaths?
Railroad fatalities do not come primarily from derailments alone. Fatalities can result from grade-crossing collisions, trespassing, train collisions, derailments and other incidents.
In the United States, people outside trains, particularly motorists and trespassers, account for a significant share of railroad-related fatalities. This is why overall rail safety statistics need to distinguish passenger accidents from the broader railroad casualty picture.
What U.S. state has the most railroad tracks?
Texas has the largest railroad network in the United States by track mileage, supported by its extensive freight-rail system and major connections between industrial, energy, agricultural and population centres.
Because track mileage and railroad activity vary substantially between states, state-by-state comparisons of derailments should account for network size rather than relying only on the number of reported incidents.