Between 2015 and 2025, trains struck vehicles and people at American railroad crossings 23,760 times. Those collisions killed 3,226 people and injured 9,093 more.

In 2025 alone, 313 people died at a grade crossing. That is the highest annual toll since 2019 and 42% above the 2020 low, according to an analysis by Ravid & Associates of every grade-crossing collision report filed with the Federal Railroad Administration.

The Rail Crossing Risk Index ranks all 50 states by how often collisions happen relative to how many crossings they actually have, and goes one level deeper than any state ranking can: it names the individual crossings struck most often in the country.

Key findings

• 3,226 people were killed at US rail crossings between 2015 and 2025, with 9,093 injured across 23,760 collisions.
• 2025 was the deadliest year since 2019, with 313 deaths.
• Driving around a lowered gate is 2.9 times more likely to be fatal than failing to stop at all. It causes fewer than half as many collisions, yet kills more people than any other driver action.
• Two in three pedestrians struck at a crossing die. Pedestrians are 8% of those hit but 41% of everyone killed.
• Arizona has the highest collision rate of any state at 24.67 per 100 crossings, nearly 26 times New Hampshire's rate of 0.96.
• California recorded the most deaths of any state, 546. Texas recorded the most collisions, 2,665.
• More than half of all collisions happen in daylight, not darkness.
• Phoenix is home to the most-struck crossing in America: Thomas Road, hit 27 times in 11 years.

A Ravid & Associates Research Report

The Rail Crossing Risk Index

Every state ranked by how often vehicles and people are struck at its railroad crossings, plus the individual crossings hit most often in America. Built from every federal grade-crossing collision report filed from 2015 through 2025.

Where collisions happen most

Shaded by collisions per 100 crossings, which compares states fairly instead of rewarding the ones with the least railroad. Click any state.

Collisions per 100 crossings Total deaths

Find your state

Search or pick a state to see its collisions, deaths and ranking.

The crossings hit most often

Individual crossings ranked by number of collisions, 2015 to 2025. Every one of these is a specific street in a specific town.

#	Crossing	City	State	Railroad	Collisions	Killed

What people were doing

Driver action in the moment of the collision, and how often each one proved fatal.

Who gets killed

Share of collisions that killed someone, by what the train hit.

Michigan in detail

Collisions by county, 2015 to 2025.

Every state

Rank	State	Per 100 crossings	Collisions	Killed	Injured	Crossings

Methodology and sources

Collision data. Federal Railroad Administration, Highway-Rail Grade Crossing Incident Data (Form FRA F 6180.57). Every impact between on-track rail equipment and a highway user at a grade crossing, as reported by railroads to the FRA. Study window is calendar years 2015 through 2025. 2026 is excluded because the year is incomplete.

Crossing counts. Federal Railroad Administration, Crossing Inventory (Form FRA F 6180.71), current file. The file holds 438,882 records, but 52,789 are grade separated (the road passes over or under the track, so a highway user never meets a train) and 196,051 are recorded as closed. This index uses only the 203,230 crossings that are both at grade and open, because those are the only ones where a collision can occur. The share of closed crossings varies by state, so this filter changes the ranking and not just the scale.

Ranking rule. States with fewer than 500 at-grade open crossings are shown in the table but left out of the ranking, because a small denominator makes the rate unstable. Those states are Alaska, Delaware, Rhode Island and the District of Columbia.

A note on the year field. The FRA incident file stores a two-digit incidentyear, so a plain sort places 1999 above 2025. Every year in this analysis is derived from the full incident date instead.

Reproducibility. Both sources are public and free, with no API key. The pull scripts and the ranking script are published with the study, so any reporter can rebuild these numbers.

Data pulled 2026-10-15.

Embed this index:

Source: Ravid & Associates, P.C., The Rail Crossing Risk Index. Analysis of Federal Railroad Administration grade-crossing collision and inventory data, 2015 to 2025.

Free to republish with attribution and a link to Ravid & Associates.

Ravid & Associates, P.C.

Trial lawyers serving Southeast Michigan

23855 Northwestern Hwy, Southfield, MI 48075

ravidandassociates.com

The gate is the dividing line

The federal data records what the driver was doing at the moment of impact, and the categories tell two completely different stories.

Failing to stop is the most common action, behind 8,198 collisions. It is dangerous: 811 people died. But it is survivable more often than not, with a fatality rate of 9.9%.

Driving around a gate that is already down is far less common, behind 3,680 collisions. And it killed 1,051 people, a fatality rate of 28.6%.

So the less frequent behaviour produced more deaths. A driver who goes around a lowered gate is nearly three times more likely to die than one who simply did not stop, and gate violations as a group account for a quarter of all crossing fatalities despite being a fifth of collisions.

The distinction matters because the two are not the same kind of event. One is a failure of attention. The other is a decision made with the warning working exactly as designed.

Who dies at crossings

A person on foot struck at a crossing dies about two times in three. Someone in a car dies about one time in twelve. Pedestrians make up a small share of those struck and a very large share of those killed.

Daylight is not the protection people assume

Over half of all crossing collisions, 53%, happen in full daylight. Darkness accounts for 28%.

More striking is that the fatality rate barely moves: 14.0% in daylight, 13.4% in the dark. Whatever is going wrong at these crossings is not mainly a matter of not being able to see the train.

The state rankings

Ranking states by raw collision count mostly ranks them by how much railroad they contain. Texas has the most collisions because Texas has the most crossings. To compare fairly, the index measures collisions per 100 crossings.

One detail matters here. The federal inventory lists 438,882 crossing records, but 52,789 of those are grade separated, meaning the road passes over or under the track and a driver never meets a train, and another 196,051 are recorded as closed. Only 203,230 crossings are both at grade and open, and that is the figure this index uses as its denominator. Using the full file would roughly halve every rate and would flatter the states that have closed the most crossings.

Highest collision rates

Lowest collision rates

The national figure is 11.69 collisions per 100 crossings. Arizona's rate is nearly 26 times New Hampshire's.

Four states and territories appear in the full table but are left out of the ranking because they have fewer than 500 at-grade open crossings, which makes a rate unstable: Alaska, Delaware, Rhode Island and the District of Columbia.

The crossings themselves

A state ranking cannot tell a driver which crossing to be careful at. The federal data can, because every collision is tagged to a specific crossing ID.

Metro Phoenix appears four times in the national top twelve. A single intersection, Thomas Road, has been the site of 27 collisions in 11 years.

Michigan

Michigan sits in the middle of the national table: 27th of the 46 ranked states, with 8.83 collisions per 100 crossings against a national figure of 11.69. Over the 11 years the state recorded 593 collisions, 61 deaths and 222 injuries across 6,718 at-grade open crossings.

The statewide number hides where the risk actually sits.

Wayne County accounts for 21% of every rail crossing collision in Michigan, 2.8 times as many as the next county. Detroit crossings that appear repeatedly in the federal record include West Chicago Boulevard, East Outer Drive, Livernois Avenue and the 7 Mile Road crossing.

Methodology

Collision data. Federal Railroad Administration, Highway-Rail Grade Crossing Incident Data, Form FRA F 6180.57, covering every reported impact between on-track rail equipment and a highway user at a grade crossing. Study window is calendar years 2015 through 2025. 2026 is excluded as incomplete.

Crossing counts. Federal Railroad Administration, Crossing Inventory, Form FRA F 6180.71, current file. The file holds 438,882 records. This index counts only the 203,230 that are both at grade and not closed, because a grade-separated crossing cannot produce a grade-crossing collision and neither can a closed one. The share of closed crossings varies by state, so filtering changes the ranking and not just the scale.

Ranking rule. States with fewer than 500 at-grade open crossings appear in the full table but are excluded from the ranking, because a small denominator produces an unstable rate.

One data note worth stating. The FRA incident file stores a two-digit year field, so a naive sort places 1999 above 2025. Every year in this analysis is derived from the full incident date.

Reproducibility. Both datasets are public, free and require no API key. The pull and ranking scripts are published with this study so any reporter can rebuild every figure.

Data pulled 2026-10-15.

Press contact

Media enquiries and data requests: Sheila Eugenio, MediaMentions, on behalf of Ravid & Associates, P.C. Full dataset, state cuts and per-crossing extracts available to journalists on request.