Fatal crash rate in San Francisco County, California
Fatal crashes recorded by NHTSA in San Francisco County, California, 2020 to 2024, divided by the people who live there. Fatal crashes only: not injuries, not all crashes, not traffic in general.
San Francisco County, California2024 · fatal crashes recorded by NHTSA
5.1fatal crashes per 100,000 residents in 2024
In 2024, NHTSA recorded 42 fatal crashes in San Francisco County, California, where 820,945 people lived. That is 5.1 fatal crashes for every 100,000 residents, and 48 people died in those crashes. Over 2020 to 2024 the average was 4.3.
The road is a scale of fatal crashes per 100,000 residents in 2024, 0 to 40. The yellow post is San Francisco County; the white posts are California and the United States.
- Per 100,000
- 42 fatal crashes ÷ 820,945 residents × 100,000 = 5.1
- Five-year
- 180 crashes (2020 to 2024) ÷ 4,142,445 resident-years × 100,000 = 4.3
- Same year
- California: 3,583 ÷ 39,364,774 = 9.1 · United States: 36,297 ÷ 340,003,797 = 10.7
Every year, 2020 to 2024
| Year | Fatal crashes | Deaths | Population | Per 100,000 |
|---|---|---|---|---|
| 2020 | 30 | 32 | 875,139 | 3.4 |
| 2021 | 33 | 34 | 816,169 | 4.0 |
| 2022 | 41 | 42 | 813,266 | 5.0 |
| 2023 | 34 | 35 | 816,926 | 4.2 |
| 2024 | 42 | 48 | 820,945 | 5.1 |
2024 is NHTSA’s Annual Report File and will be revised when the final file replaces it. Populations are July 1 estimates, Census Vintage 2025.
What the record says about those crashes, 2020 to 2024
Every figure is a count of the 180 fatal crashes NHTSA coded to San Francisco County, California over five years, from the fields FARS records. Nothing is estimated or inferred.
| Road type | Crashes | Share |
|---|---|---|
| Interstate | 19 | 11% |
| Other freeway or expressway | 14 | 8% |
| Other principal arterial | 75 | 42% |
| Minor arterial | 33 | 18% |
| Major collector | 14 | 8% |
| Minor collector | 1 | 1% |
| Local road | 24 | 13% |
| Unknown or not reported | 0 | 0% |
| Rural | 2 | 1% |
| Urban | 178 | 99% |
| Time of day | Crashes | Share |
|---|---|---|
| Midnight to 5:59 am | 57 | 32% |
| 6 am to 11:59 am | 31 | 17% |
| Noon to 5:59 pm | 39 | 22% |
| 6 pm to 11:59 pm | 51 | 28% |
| Hour unknown | 2 | 1% |
| Light | Crashes | Share |
|---|---|---|
| Daylight | 71 | 39% |
| Dark, not lighted (or lighting unknown) | 16 | 9% |
| Dark, lighted | 87 | 48% |
| Dawn or dusk | 6 | 3% |
| Unknown or not reported | 0 | 0% |
| Weather | Crashes | Share |
|---|---|---|
| Clear | 151 | 84% |
| Cloudy | 19 | 11% |
| Rain | 6 | 3% |
| Snow, sleet or freezing rain | 1 | 1% |
| Fog, smog or smoke | 1 | 1% |
| Severe wind, blowing sand or other | 0 | 0% |
| Unknown or not reported | 2 | 1% |
| Involving | Crashes | Share |
|---|---|---|
| A pedestrian | 85 | 47% |
| A bicyclist or other cyclist | 5 | 3% |
| A speeding-related vehicle | 67 | 37% |
| A driver police reported as drinking | 21 | 12% |
| Most-named roads | Crashes |
|---|---|
| US-101 | 11 |
| I-80 | 11 |
| 16TH ST | 11 |
| I-280 | 7 |
| MISSION ST | 7 |
Road type is FARS FUNC_SYS; speeding is any vehicle coded SPEEDREL 2 to 5; drinking is any driver coded DRINKING = 1 (police-reported, not a blood-alcohol test). Road names are the trafficway strings as the police coded them, counted as written, so the same road can appear under two spellings.
How this is counted
- Fatal crashes are rows in NHTSA’s Fatality Analysis Reporting System (FARS) file
FARS2024NationalCSV.zip,accident.csv, one row per crash on a public road in which someone died within 30 days. Deaths are theFATALSfield. Counties are the FARSSTATEandCOUNTYcodes joined to Census FIPS codes. - Population is the Census Bureau’s July 1 estimate for the same year, county totals, Vintage 2025. Rate per 100,000 = crashes ÷ population × 100,000.
- The five-year average is total crashes over the years with both inputs ÷ total population over those years × 100,000, the rolling form in 23 CFR 490.207.
- What it cannot say. FARS has no injury-only or property-damage crashes, no congestion, and nothing about any one driver, trip or road’s risk. A crash is counted where it happened, not where the people lived. Rates are not ranked here; comparing two places means comparing both sets of inputs.
Informational only, not professional advice. Last reviewed: October 2026. Data snapshot built 2026-10-28.