The Deadliest Highway in Every State

How to read these numbers

The state table ranks highways by deaths. A highway that runs the length of a large state has more room for crashes than a short one, which is why interstates lead in most states. The per-mile table corrects for length. It does not correct for traffic, because no consistent national count of vehicles per highway exists in these files.

The records cover fatal crashes only. They include people walking and cycling as well as people in vehicles. A crash that injured someone without killing anyone is not in the data, and neither is a road that had no fatal crash on record during these three years.

A highway's count depends on the road name recorded in the crash file. Where two routes share the same pavement, the crash is counted under the primary road name in the record. Interstate branches such as I-35E and I-35W are counted with I-35. Lettered U.S. and state routes, such as US-31W in Kentucky, SR-99E in Oregon and New York's numbered parkways, are separately signed highways and are counted on their own. A record that gives only the parent number stays with the parent. Missouri's lettered state routes do not follow the numbered pattern and are left out.

Texas in more detail

Texas has its own page with a map of fatal crashes on 291 highways, county breakdowns and yearly totals: Most Dangerous Highways in Texas. For commercial-vehicle crashes in each Texas county, see Texas truck accident statistics by county. That page uses a different source and its totals should not be combined with the figures here.

Open methodology

How we counted the crashes

We used the national 2022, 2023 and 2024 NHTSA Fatality Analysis Reporting System (FARS) CSV files, downloaded September 18, 2026. NHTSA's FARS definition requires a motor vehicle traveling on a trafficway customarily open to the public and a death within 30 days. We do not claim that 2024 is final.

1. Start with every fatal crash. The three files hold 113,488 fatal-crash records for the 50 states and the District of Columbia.
2. Match the numbered highway. Keep crashes coded to the Interstate, U.S. highway or state highway system whose primary road name starts with a matching I-, US- or SR- number. That leaves 61,842 crashes on 8,288 highways, counting each state's stretch of a route separately. Another 48,955 crashes happened on county roads, city streets, unsigned roads and other road types.
3. Leave out ramps and frontage roads. 1,964 records name a ramp, frontage road, business route, connector or spur, or are coded to an entrance or exit ramp. 727 records have a road name that does not fit the numbered pattern.
4. Flag large trucks. Join the FARS vehicle table to each crash. A crash counts as involving a large truck when any vehicle has a body type that FARS describes as a truck above 10,000 pounds gross vehicle weight rating (codes 60 to 64, 66, 67, 71, 72 and 78). This follows NHTSA's large-truck definition, which leaves out buses and motor homes. We also leave out code 79, trucks of unknown weight, so our counts can run slightly below NHTSA's published totals.
5. Measure the interstates. Interstate miles in each state come from FHWA's Interstate Route Log and Finder List, main and auxiliary route tables, as of January 2026. Deaths per 100 miles is the three-year death total divided by those miles. We rank interstates with at least 25 miles in the state and 15 deaths. We do not compute a rate for U.S. or state highways.

One caution on the per-mile figures: FHWA counts only the miles signed as an interstate. Where a freeway keeps the same number past the end of its interstate designation, crashes on the extra miles can be recorded under the interstate's name, which pushes the rate up.