The Crash Report Sampling System (CRSS) estimates how many motor vehicle traffic crashes are reported to police in the United States. Use it for national estimates of injury and property-damage-only crashes. For a national fatality count, use the Fatality Analysis Reporting System (FARS). NHTSA overview of 2024 motor vehicle traffic crashes, Data Sources

Which crashes can enter the sample?

CRSS starts with official crash reports signed by law enforcement. To qualify, a report must describe a crash involving at least one motor vehicle traveling on a trafficway and resulting in damage, injury or death. A collision that nobody reports to police falls outside the population CRSS estimates. That matters when comparing a CRSS estimate with a survey that asks people directly about crashes they experienced. NHTSA Crash Report Sampling System, Sampling

The reports come from sampled areas and police jurisdictions. Staff convert the selected reports to a common set of variables. Information has to come from the selected reports, so what officers record affects the detail available for analysis. For some variables, NHTSA uses statistical methods to fill missing values. If a count depends on a particular detail, check the documentation for whether it uses those filled-in values. NHTSA Crash Report Sampling System, Data Files NHTSA CRSS design guidance, sections 2 and 6

Why a row count differs from a national estimate

Some reports have a different chance of selection from others. CRSS assigns weights so that the sampled reports can represent the larger population. A filter might return hundreds of records in the public file while the weighted estimate represents many more crashes nationally. The number of records still matters when assessing precision, so read any reported uncertainty or reliability note alongside a small subgroup’s estimate. NHTSA CRSS design guidance, sections 3–6

For a question about one intersection, use crash records from the agency responsible for that location. Sampling jurisdictions around the country supports the national estimate. It gives you little reason to expect complete records for any individual road. NHTSA CRSS design guidance, section 2

How to trace a change across releases

Keep the estimate’s source and method beside the result. NHTSA introduced CRSS in 2016 as the replacement for the General Estimates System (GES). Its crash overview says the two systems use different sample designs and advises against comparing their estimates directly. For a CRSS trend, pick comparable years within the series and read the significance note. In the 2024 crash overview, the estimated national crash total rose from 6.14 million to 6.18 million, a change NHTSA reported as statistically inconclusive. NHTSA overview of 2024 motor vehicle traffic crashes, summary and Data Sources