Geographic Risk: Your City's AV Safety Score
Autonomous vehicle risk isn't distributed evenly. San Francisco has 1,170 incidents; most American cities have zero. Geography determines what kind of AV you'll encounter and how likely it is to be involved in a crash.
Key Insights
- →San Francisco leads with 1,170 incidents — 19% of the national total from a single city
- →The top 5 cities account for over 60% of all AV/ADAS incidents nationwide
- →Waymo-heavy cities (SF, Phoenix) have high incident counts but very low fatality rates
- →Tesla-heavy cities show fewer total incidents but higher severity and fatality rates
- →53 states have reported AV/ADAS incidents; California alone has 44% of the total
1,193
SF Incidents (#1)
1274
Cities with Incidents
53
States Reporting
44%
California's Share
San Francisco: Ground Zero for Autonomous Driving
With 1,170 reported incidents, San Francisco isn't just the #1 city for AV crashes — it's in a league of its own. The next closest city, Los Angeles, has roughly a third of that count. San Francisco's dominance comes from one factor above all others: Waymo's massive robotaxi fleet. Over 1,100 of SF's incidents are ADS (fully autonomous), reflecting thousands of Waymo Jaguar I-PACEs navigating the city's steep hills, dense traffic, and chaotic streets every day.
But here's the nuance: despite 1,170 incidents, San Francisco has recorded just 1 fatalityfrom AV operations. That's a fatality rate of 0.09% per incident. The overwhelming majority of SF's AV crashes are low-speed fender benders — a Waymo getting rear-ended at a stoplight, a minor sideswipe during a lane change, a pedestrian clipping a mirror at walking speed. High volume, low severity.
Phoenix: The AV Testing Ground
Phoenix and its suburbs form the second major AV testing cluster. Waymo has operated in Chandler, Tempe, and Mesa since 2017, and its presence shows in the data. Phoenix's wide, grid-patterned roads and predictable weather make it an ideal AV environment — and Waymo's serious injury rate there is just 0.01 per million miles, the lowest of any city in their data.
The Phoenix metro also has historical significance: it's where Uber's self-driving car struck and killed Elaine Herzberg in Tempe in 2018, the first pedestrian fatality involving an autonomous vehicle. That tragedy predates the SGO database but looms over all AV operations in the area.
Los Angeles and Austin: Waymo's Expansion Cities
Los Angeles and Austin represent Waymo's newest major markets. LA has logged hundreds of incidents as Waymo scales up service. Austin — one of Waymo's fastest-growing markets — is generating a similar pattern: lots of minor urban incidents, very few serious injuries. Both cities show 0.00 serious injuries per million miles from Waymo so far, though the sample sizes are smaller than SF and Phoenix.
The Tesla Geography: Everywhere, with Higher Stakes
While Waymo concentrates in a handful of cities, Tesla's incidents are distributed across 51 states and territories— because Tesla is a consumer car company, not a robotaxi service. You won't find a concentrated cluster of Tesla ADAS incidents in one city. Instead, they're scattered along highways and suburban roads from Maine to Hawaii.
This distribution matters for risk assessment. In cities where both Waymo and Tesla operate (like LA, SF, and Austin), the incident profiles are starkly different: Waymo generates many low-severity ADS incidents, while Tesla generates fewer but more severe ADAS incidents. A city's "AV risk score" depends heavily on which AV technology is prevalent there.
The State-Level Picture
California dominates with 2,838 incidents (44% of the total), followed by Texas, Arizona, and Florida. But the fatality rate tells a different story: Florida, with far fewer incidents, has a disproportionately high fatality count — suggesting that its incidents tend to be high-speed highway crashes rather than urban fender benders.
States without AV-specific legislation (like many in the Midwest and Southeast) have far fewer incidents — not because they're safer, but because AV companies haven't deployed there yet. As Waymo expands to Atlanta, Miami, and other cities, expect the geographic distribution to shift significantly.
What Drives City Risk Differences
Three factors determine a city's AV safety profile: which AV companies operate there (Waymo = high volume, low severity; Tesla = lower volume, higher severity), urban density (denser cities generate more low-speed contacts), and road design (grid patterns with wide lanes favor AV performance; narrow, winding streets with poor markings stress AV sensors).
Cities investing in AV-friendly infrastructure — clear lane markings, dedicated AV zones, V2X communication systems — may see lower incident rates as the technology matures. But for now, geography is a strong predictor of both the quantity and quality of AV crash risk you'll encounter.
Top 20 Cities by AV/ADAS Incidents
Red bars indicate cities with fatalities.