How We Count

Methodology

injured.org publishes findings built from federal crash and injury records. Not marketing copy. Not talking points. Every number on this site traces to a named public source, a set date range, and a documented method. This page explains how we build them.

Our primary source

Every death count on this site starts with the Fatality Analysis Reporting System (FARS). FARS comes from NHTSA, the National Highway Traffic Safety Administration. NHTSA is part of the U.S. Department of Transportation. It is a census of every fatal motor vehicle crash on a U.S. public road. That means a full count, not a survey or a sample. NHTSA builds a record of every crash that killed someone within 30 days. It draws on police reports, coroner and medical examiner records, and state crash files. We use FARS data from 2000 through 2024, the most recent year NHTSA has finalized. NHTSA usually finalizes a year's data about two years after it ends. Our numbers update on that same yearly schedule.

How we estimate injuries

FARS only counts deaths. To estimate injuries, we use NHTSA's Crash Report Sampling System (CRSS). CRSS is not a full count. It is a sample of police-reported crashes of every severity. The sample is built to stand in for the whole country. CRSS scales its sample up to a national estimate. So every injury number on this site is labeled an estimate and links back to this page. CRSS estimates about 54 people are injured for every person killed in a crash. That covers recent years. The ratio is a national estimate only. CRSS can't support state or county injury numbers. So we never present a CRSS-based number as a state or local statistic.

Rates for every 100,000 residents

Raw death counts favor big states. To compare states and counties fairly, we adjust for population. We report each rate as deaths for every 100,000 residents. Population comes from U.S. Census Bureau estimates for the same year as the crash data. A rate always uses the latest population number available for that place.

The 2009 hit-and-run coding change

NHTSA changed how it labels hit-and-run crashes over time. From 2009 through 2019 it used a separate "unknown" label. It retired that label later. So every hit-and-run trend or comparison here uses 2009-2024 data only. We never plot hit-and-run numbers against years before 2009. The labels don't match across years that far back.

What we don't claim

A dataset can't answer every question. Some numbers would need us to guess at a cause NHTSA doesn't record. Some would rank small counties on a single year's rate. Some would combine two datasets that measure different things. In each case we say so directly instead of publishing a number we can't stand behind.

Corrections and updates

If you find an error in a number, a broken source link, or a method we've described wrong, reach us through our about page. We correct published numbers as soon as we confirm the fix. We note meaningful changes on the page they affect. See our editorial policy for the full correction process. Our data sources page shows where every dataset comes from.