Drug incidents: 2010 → 2026

SFPD incident reports, Jan. 1–Sep. 23 in each year
6,9422010 Drug/Narcotic records
7,1852026 Drug Offense records
7.7%share of all 2010 report rows
10.7%share of all 2026 report rows
2010 drug share7.7%
2026 drug share10.7%
What changed?
Drug-related report rows are slightly higher in 2026 (+243, +3.5%), even though total incident-report rows fell from 90,192 to 66,862. As a share of all report rows, drug records rose from 7.7% to 10.7% — an increase of 3.0 percentage points.

The composition changed too

In 2010, leading descriptions included narcotics paraphernalia possession, marijuana possession, base/rock cocaine possession and sales, and methamphetamine possession.

In 2026, Narcotics Paraphernalia, Possession of accounts for 3,953 drug records by itself. Other leading descriptions include methamphetamine offenses (498), controlled-substance intoxication (442), loitering where narcotics are sold/used (325), and opiate offenses (259).

The category systems are not identical. “Drug/Narcotic” (2010) and “Drug Offense” (2026) should therefore be treated as closely related reporting categories, not a perfectly unchanged legal definition.

Map: housing & harm reduction

This version intentionally removes the police-call, 311, HEART, RESET and other operational layers. It retains the housing/service-site geography and the focused harm-reduction locations from the earlier map so the historical reporting comparison can be read alongside the present-day service landscape.

Permanent supportive housing
Shelter / navigation / temporary housing
Other HSH/DPH housing or service site
Harm-reduction focus site
2010 Drug/Narcotic incidents
2026 Drug Offense incidents

Use the layer control in the upper-right of the map to show or hide categories. The incident layers contain 6,942 geocoded 2010 drug-report rows and 7,127 geocoded 2026 drug-report rows for Jan. 1–Sep. 23. They are off by default so the housing and harm-reduction geography remains readable. Click a marker for site details. Map points describe the service-site dataset used in the prior project; they do not establish that a particular site caused or prevented nearby incidents.