Where federal crash data comes from: FARS, CRSS and how to use them
How the government counts crash deaths and injuries and how to read the numbers
When a news story says “about 37,000 people died on U.S. roads last year,” where does that number come from? Most of the time, it comes from the National Highway Traffic Safety Administration (NHTSA), which runs the nation’s main crash data systems.
Knowing how these systems work helps you read the news, write a grant, make the case for a new crosswalk or simply understand the risks you face. This guide explains each major source in plain English.
The big picture: what we know right now
NHTSA’s early estimate for 2025 counts 36,640 traffic deaths, a drop of about 6.7 percent from 39,254 in 2024 [1]. The death rate was 1.10 per 100 million vehicle miles traveled. If it holds, that would be the second lowest rate on record, behind 1.08 in 2014 [1].
Early 2026 trends point slightly lower. The National Safety Council’s preliminary estimate for the first half of 2026 was 18,460 deaths, down 2 percent from the same period in 2025 [2].
FARS: every fatal crash
The Fatality Analysis Reporting System, or FARS, is a census of fatal traffic crashes. It includes every crash on a public road in the 50 states, Washington, D.C., and Puerto Rico in which someone died within 30 days [3].
Trained analysts in each state pull information from police crash reports, death certificates, vehicle registration files, driver records, highway data, hospital and EMS records, and toxicology results. FARS records more than 100 details per crash, such as the time, road type, weather, vehicles, seat belt use and alcohol results.
CRSS and CISS: samples of all crashes
Most crashes do not kill anyone, so NHTSA estimates them using samples [3]:
- CRSS, the Crash Report Sampling System, started in 2016. It draws a national sample of police reported crashes of all severities. It replaced the older General Estimates System.
- CISS, the Crash Investigation Sampling System, also started in 2016. It sends investigators to study a sample of crashes where a passenger vehicle was towed. It replaced the older Crashworthiness Data System.
- Special Crash Investigations take a deep look at about 200 unusual cases a year, often involving new vehicle technology [3].
- Non Traffic Surveillance counts deaths and injuries off public roads, such as in driveways and parking lots [3].
Comparison compiled by Safety Behind the Wheel Foundation
NHTSA’s main crash data systems
| System | What it covers | How | Since |
|---|---|---|---|
| FARS | All fatal crashes on public roads | Census | 1975 |
| CRSS | Police reported crashes of all severities | National sample | 2016 |
| CISS | Crashes with a towed passenger vehicle | National sample with in depth investigation | 2016 |
| Special Crash Investigations | Unusual or high interest crashes | Case by case, about 200 a year | 1972 |
Source: NHTSA National Center for Statistics and Analysis [3].
Why the numbers keep changing
Crash data arrive in stages:
1. Early estimates come out every few months, based on quick counts from states [1].
2. The Annual Report File follows about a year later.
3. The Final File comes about two years after the year ends. NHTSA expects the 2024 final FARS file and the 2025 annual file in early 2027 [1].
Totals usually change only a little between stages. When you cite a number, say which year and which version it is.
Why different sources disagree
You may see different death totals for the same year. That is usually not an error.
- NHTSA counts traffic deaths on public roads within 30 days of the crash.
- The National Safety Council uses death certificate data and includes some deaths that NHTSA does not, such as non traffic deaths in parking lots and driveways.
- CDC data come from death certificates and use different categories.
For example, for people 65 and older in 2024, NHTSA counted 8,019 traffic deaths while CDC counted 9,727 motor vehicle deaths [4]. Both are accurate for what they measure.
Limits to keep in mind
- Police reports vary. Officers record what they can see at the scene. Some factors, such as drowsiness, distraction and drug use, are hard to prove and often undercounted.
- Missing data. Some details, like whether a phone was in use, are often unknown.
- Small numbers swing. In a small town or county, one bad crash can make a year look much worse than average.
How to look up federal crash data
- NHTSA CrashStats offers free reports and fact sheets by topic, such as older drivers, pedestrians and alcohol [3].
- The Fatality and Injury Reporting System Tool (FIRST) lets you build your own tables from FARS and CRSS.
- State Traffic Safety Information pages summarize FARS data by state and county.
Checklist: citing crash data well
Checklist from Safety Behind the Wheel Foundation
- I named the source, such as FARS, CRSS or a state database
- I gave the year and said whether it is an early estimate or final
- I used rates per population or per mile when comparing places
- I noted limits, such as undercounted causes
- I linked to the original report
Frequently asked questions
How long after a crash does a death count in FARS?
The person must die within 30 days of the crash for it to count as a traffic fatality in FARS.
Can I find crashes on my street in FARS?
FARS includes location details for fatal crashes. For all crashes, including injury and property damage crashes, your state’s crash data portal is usually better.
Why do news reports use “crash” instead of “accident”?
Safety agencies and many newsrooms prefer “crash” because most crashes involve preventable factors, such as speeding, impairment or road design.
Related articles
Sources and further reading
- 1National Highway Traffic Safety Administration. “Early Estimate of Motor Vehicle Traffic Fatalities in 2025” (DOT HS 813 800). April 2026. crashstats.nhtsa.dot.gov/…/813800
- 2National Safety Council. “Traffic Fatalities Down Slightly in First Half of 2026, Preliminary NSC Analysis Shows.” nsc.org/…/traffic-fatalities-down-sli…
- 3National Highway Traffic Safety Administration, National Center for Statistics and Analysis. “NCSA and Other Data Sources.” cdan.dot.gov/…/MotorVehicleCrashDataOvervi…
- 4National Safety Council, Injury Facts. “Older Drivers.” injuryfacts.nsc.org
This article reflects the safety priorities of traffic safety data and public education. It is general education. Always check the original data source for the latest figures. Reviewed November 2026.
Special thanks to our contributors
We appreciate the professionals, experts, and community members who share their knowledge and perspectives to help make this resource better. Your insights are always welcome and appreciated.
Have professional expertise, firsthand experience, or a perspective that could make this resource better?
We welcome your input. Your knowledge and experience can help us make this resource more useful, accurate, and meaningful to the people who rely on it.
Thank you. Your contribution is on its way.
We’re grateful you took the time to share your knowledge with us. Our team reviews every contribution with care. If it helps improve this resource, we’ll update the page and recognize you the way you chose.
Keep learning
Safe Driving Resource Center
Guides, self-checks and articles for drivers, parents of new drivers, and families recovering after a crash.
