A global-scale research estimate anchored by official 2025 FBI city-agency evidence, with United States-level fallbacks and explicit uncertainty for dimensions without city coverage.
Safrom's citywide research estimate is 75/100. Higher scores mean safer on the model's scale. Evidence as of . Confidence: C.
What does this area's map cover?
11 areas with estimates. Area evidence as of .
Area evidence and limitations
Overall-only public-alpha estimate. Supervisor-district variation uses selected person and property incident categories in the San Francisco Police Department's 2025 Incident Reports dataset, grouped by the official supervisor_district field and conservatively anchored to San Francisco's city score. The official DataSF rows were transported through a read-only text proxy after the source host failed TLS from the build network. Counts are selected incident-category records; one incident can carry more than one category. They are not resident or visitor victimization rates, and district population, commuters, tourism, reporting, policing and land use materially affect the pattern. No day/night differences are inferred. Official supervisor-district boundaries are for display; not travel advice.
Locate your intended stay on the map and check whether that area has evidence.
Read what drives the local differences. A citywide score and an area estimate answer different questions.
Check current local advice and your transport options separately. These estimates cannot assess a specific hotel, street or time of day.
Citywide evidence
Personal violence
66/100
Property & scams
58/100
Getting around
80/100
Health & rescue
88/100
Conflict & stability
86/100
Disaster & environment
78/100
Infrastructure response
88/100
Batch-generated global-scale research estimates. Local signals are used when available; otherwise documented country-level baselines and official current-condition guidance provide a low-confidence first reference. Not travel advice.
the repository's staged 2025 snapshot is preliminary and blocked after the final annual release was detected; FBI warns against simplistic city rankings; reported offenses are not traveler exposure; law-enforcement agency coverage may differ from municipal boundaries; Florida and North Dakota have limited 2025 data and are excluded from the first cohort
country-level only; modeled perceptions rather than traveler incidents; cannot represent neighborhood variation or current events; confidence intervals must not be discarded