All sources · source collection

Table FI-200: Motor Vehicle Traffic Fatalities, 1900-2023 (Highway Statistics 2023)

Federal Highway Administration, Office of Highway Policy Information · 2025

Population measured

All persons killed in motor-vehicle traffic crashes on US public roads (the States and the District of Columbia), measured against total vehicle-miles travelled — not drivers only and not vehicle occupants only

Read the original source ↗

Source notes & limitations

The whole 1900-2023 span in one FHWA table, so the fatality-rate series needs no splice between estimate and registration eras. Four rate columns are published side by side (per 1,000 miles of road, per 100 million annual VMT, per 100,000 registered motor vehicles, per 100,000 licensed drivers); the museum takes the per-100-million-VMT column, and the extraction script recomputes it from the table's own fatality and VMT columns for every year rather than trusting its position. FHWA footnote (3) records a definitional change: beginning in 1976 the count includes only people who died within 30 days of the crash, so the earlier years are counted on a wider window and the two halves are not the same measurement. The 1900-1995 rows were cross-checked against FHWA's separately-published Highway Statistics Summary to 1995 (April 1997): 95 of 96 years reproduce exactly, the exception being 1995, that publication's terminal provisional year, since revised from 41,770 deaths to 41,817 (see docs/verification-log.md).

Where this source appears

13 records across 13 decades and countries. Documented gaps are included; a source cited for a gap does not establish a value.

0 of 13 records have a current source audit. A transcription check reads a declared numeric field from a pinned local source file; a calculation replay reads its operands and applies the declared method and rounding rule. A source audit does not independently validate the source or every claim in a record. The date is the last successful local audit; the build checks that its bindings still match.