An 18-month review required by state law found that all 33 of the city’s camera sites passed California’s performance test, though officials have not yet measured the effect on crashes.
By Staff Reporter
San Francisco’s automated speed cameras have sharply reduced dangerous speeding where they operate, according to an 18-month evaluation released Thursday by the San Francisco Municipal Transportation Agency, the first such review under California’s speed camera pilot law.
The share of drivers traveling 10 miles per hour or more over the posted limit fell by close to 84 percent across the city’s 33 camera locations compared with pre-installation levels, the agency found, or an estimated 42,000 fewer instances of excessive speeding each day. Mayor Daniel Lurie announced the results on Thursday as part of his Street Safety Initiative.
“My top priority is public safety, and that includes making our streets safer for everyone who uses them,” Mr. Lurie said in a statement.
San Francisco was the first of six cities authorized by Assembly Bill 645, signed in October 2023, to switch on cameras, making its data the first full test of the benchmarks the Legislature wrote into the law.
What the numbers show
The SFMTA operates 56 cameras at 33 locations, all on the city’s High Injury Network, the small fraction of streets where most severe and fatal collisions occur. The first cameras were activated on March 20, 2025, and citations began on Aug. 5, 2025.
Owners are cited for vehicles recorded going 11 m.p.h. or more over the limit. Fines start at $50, with discounts for low-income owners, and the violations are civil, carrying no points on a driver’s license.
On Geneva Avenue between Prague Street and Brookdale Avenue, a 30 m.p.h. street, the agency recorded 6,630 vehicles going at least 10 m.p.h. over the limit in a pre-installation traffic count, about 31 percent of all traffic. In a May 2026 count, it recorded 144, or about 1 percent. The speed at or below which 85 percent of drivers traveled on that block fell from 43 m.p.h. to 32 m.p.h.
The steepest declines, 97 to 99 percent, were on Monterey Boulevard, Geneva Avenue and Market Street at Danvers Street. The smallest, near 60 percent, were on Guerrero Street in the Mission and on Fulton Street near Arguello Boulevard. Near schools and senior centers, excessive speeding fell by as much as 94 percent.
Most drivers who were caught appear to have adjusted. Nearly 63 percent of vehicle owners who received a notice did not receive a second, and about 81 percent received two or fewer. The share of citations issued to drivers going 16 m.p.h. or more over the limit dropped from more than half at launch to less than 3 percent. The volume of enforcement has still been substantial: 588,124 warnings since March 2025 and 235,792 notices of violation since August 2025.
A test written into state law
AB 645 requires each camera location to meet at least one of three tests after 18 months: a lower 85th-percentile speed, a 20 percent drop in drivers going 10 m.p.h. or more over the limit, or a 20 percent drop in repeat violators. A camera at a location that fails all three must be moved or shut off unless the city adds physical traffic calming.
All 33 San Francisco locations passed, chiefly on the severe-speeding measure. The passing grade masks some variation: at the outer Mission Street camera near Ottawa Avenue, the 85th-percentile speed rose slightly, and at the Fulton Street site near Ocean Beach, the repeat-violator measure moved in the wrong direction, even as severe speeding at both fell by roughly 90 percent.
A preliminary, independent analysis by the Safe Transportation Research and Education Center at the University of California, Berkeley, found that speeding dropped significantly during the warning period, before any fines were issued, and remained well below baseline 40 to 50 weeks after activation. Speeding also declined on comparison streets without cameras, which the researchers said may reflect a broader shift in behavior.
“Our finding that speeding reduced substantially during the warning period suggests that public awareness and warning notices encouraged safer driver behavior — even before fines were introduced,” said Julia Griswold, the center’s director.
The question still unanswered
The report does not yet show whether the cameras have reduced crashes, injuries or deaths, the outcomes the program exists to prevent. The agency wrote that “as information becomes more available,” it will analyze collision data.
The citywide trend has been favorable. San Francisco recorded 25 traffic deaths in 2025, down from 43 in 2024 and the lowest total since 2018, according to the SFMTA. But no published study has isolated the cameras’ share of that decline.
The report puts the program’s costs from November 2023 through June 2026 at about $4.3 million, including a $1.5 million-a-year vendor contract. First-year citation revenue was $6.5 million, leaving about $2.2 million that state law requires the city to spend on traffic calming, such as speed humps and flashing crosswalk beacons.
A statewide pilot with an end date
Other Bay Area cities are following. Oakland began issuing warnings from 35 cameras at 18 locations on Jan. 14, 2026, and started citations on March 15, according to an Oakland Department of Transportation presentation.
The pilot is temporary: the authorization expires on Jan. 1, 2032, and each city’s camera count is capped in proportion to its population. The SFMTA wrote that it will work with the other pilot cities to press the Legislature to extend the program, allow more locations and refine the rules. Supervisor Matt Dorsey introduced a resolution in July reaffirming the Board of Supervisors’ support.
The current law also sets strict privacy rules: cameras photograph only a vehicle’s rear license plate, facial recognition is prohibited, and images not tied to a violation must be deleted within five days. San Francisco’s next task is to show lawmakers that slower traffic has also meant fewer people hurt.
