SFPD incident-record file
EXPLORE 41 SAN FRANCISCO ANALYSIS NEIGHBORHOODS
Click San Francisco.
Learn the local pattern.
Choose an SFPD incident category, then select an analysis neighborhood to see raw published incident-record counts and local context.
Records through August 29, 2026 · hourly checks active · 60,046 incident records reviewedSan Francisco. Citywide view. All published incident records.
Drawing San Francisco streets and analysis neighborhoods…
Raw published count
Raw published count
Raw published count
SAN FRANCISCO AT A GLANCE
Read the local pattern.
DataSF publishes one row per SFPD incident record after supervisory review. Reports may be filed by officers or submitted by members of the public, and the counts can change as police records are updated.
01 / ANALYSIS NEIGHBORHOODS
Highest raw counts
- 01Mission9,244
- 02Tenderloin9,070
- 03South of Market7,703
- 04Bayview Hunters Point3,875
- 05Financial District/South Beach2,679
- 06Nob Hill1,582
- 07Western Addition1,571
- 08Castro/Upper Market1,401
- 09Sunset/Parkside1,372
- 10Hayes Valley1,279
- 11Marina1,204
- 12Chinatown1,087
Raw totals are not adjusted for residents, visitors, land use, or area size.
02 / CATEGORY MIX
Leading published groups
Category totals reflect the source’s own record and classification model. They should not be silently compared with another city’s local labels.
PEOPLE IN THE PUBLISHED RECORD
Who the source describes.
San Francisco publishes usable demographic fields for the profiles below. Each percentage uses only the denominator printed beside that lens.
Compare demographic lenses across cities →23,145
aggregated person countsQuarterly SFPD victim demographic totals for the reportable person-crime categories in the city dataset.
23,145 coded rows
DataSF publishes combined race and ethnicity categories for this mandated report.
23,145 coded rows
Nonbinary, unknown, and the source’s “Others” category remain visible.
23,145 coded rows
Source-defined quarterly age bands are preserved.
Coverage boundary. These are quarterly aggregates, not person-level rows. The source’s “Others” category stays visible because it can represent source-specific reporting conditions.
Records through Q2 2026 · data as of July 28, 2026.
SFPD Victim and Suspect Demographic Data ↗14,968
aggregated person countsQuarterly SFPD suspect demographic totals for the reportable person-crime categories in the city dataset.
14,968 coded rows
DataSF publishes combined race and ethnicity categories for this mandated report.
14,968 coded rows
Nonbinary, unknown, and the source’s “Others” category remain visible.
14,968 coded rows
Source-defined quarterly age bands are preserved.
Coverage boundary. A suspect record is not a finding of guilt. High unknown counts remain visible and make suspect comparisons especially limited.
Records through Q2 2026 · data as of July 28, 2026.
SFPD Victim and Suspect Demographic Data ↗READ BEFORE USING
What these numbers mean.
This dashboard reads an official public police dataset and checks it hourly. Records can change as reports are reviewed, corrected, added, or removed.
Records, geography & refresh
Record model. Each row is a published incident record, not necessarily a unique event, victim, arrest, or conviction. The source can contain multiple categorized records connected with one incident number.
Map geography. The map uses DataSF’s 41 analysis neighborhoods. They are consistent reporting groupings—not codified neighborhood boundaries—and the totals are not adjusted for population.
Refresh. DataSF says the incident file is updated automatically each day by 10:00 a.m. Pacific. This site checks the source hourly and displays the latest occurrence date returned.
Interpretation. A police record is not a conviction, and a raw count is not a population-adjusted crime rate. Use the map to examine patterns inside one consistent source.