Waymo has updated its Safety Impact dashboard with figures through June 2026. The fleet has now covered 271.3 million rider-only miles — miles driven with nobody behind the wheel — and the company's own analysis puts its injury-crash rate 82% below a human benchmark.
The numbers
Waymo compares itself against police-reported crash rates for human drivers in the cities it operates — Phoenix, San Francisco, Los Angeles, Austin and Atlanta — adjusted for how driving is distributed locally.
| Crash type | Reduction | Crashes avoided |
|---|---|---|
| Any-injury crash | 82% | 841 |
| Serious injury or worse | 95% | 55 |
| Pedestrian injury crash | 93% | 86 |
| Cyclist injury crash | 86% | 56 |
| Airbag deployment, any vehicle | 82% | 358 |
The pedestrian and cyclist rows are the ones that translate best to European cities, where vulnerable road users account for a far larger share of road deaths than they do in Phoenix.
The mileage is compounding
The 271.3 million figure covers operations through June. The previous update, through the end of March, stood at roughly 221 million. Waymo added about 50 million driverless miles in a single quarter.
That rate is the context for the comparison Tesla keeps inviting. When Tesla's VP of AI announced on 3 September that the robotaxi programme had passed 1 million unsupervised miles, Waymo's comparable total was 220 million. At its current pace Waymo adds Tesla's entire cumulative unsupervised mileage roughly every two days.
Self-reported is not independent
These are Waymo's figures, computed by Waymo, against a benchmark Waymo constructed. That is not a reason to dismiss them, but it is a reason to keep them separate from outside work.
An independent check does exist. In August the US Insurance Institute for Highway Safety published an analysis of about 80 million kilometres of driverless Waymo operation and found 68% fewer police-reportable crashes and 81% fewer injury crashes than human drivers. The injury figure is close to Waymo's own, on a much smaller sample — but the study also found a wide spread between cities, with Austin coming out 4% worse than human drivers.
Why this matters in Europe
No Waymo vehicle carries passengers in Europe yet. The company has named Munich as its first European city for 2027 and has 50 test vehicles in Berlin.
The immediate relevance is regulatory. Europe is deciding right now what evidence is sufficient to approve an automated driving system, and the vote on Tesla's Article 39 request has just slipped to December. Tesla's file rests on roughly 3 billion kilometres of supervised driving with a human responsible throughout. Waymo's rests on 271 million miles with nobody responsible at all. Those are different claims about different systems, and European regulators have not yet said which kind of evidence they find more persuasive.