Waymo Says Self-Driving Cars Cut Injury Crashes by Over 80 Percent

TLDR

• Waymo released new safety data covering 270 million fully autonomous miles through June 2026.
• The data spans five U.S. cities: Atlanta, Austin, Los Angeles, Phoenix, and San Francisco.
• Waymo says its vehicles had 82% fewer injury-causing crashes than human drivers over the same distance.
• The company reports 95% fewer crashes involving serious injury or worse.
• The Insurance Institute for Highway Safety praised Waymo for publishing the data openly.

---

Waymo has released a new set of safety figures covering its self-driving ride-hailing service. The data covers 270 million fully autonomous miles driven through June 2026.

270M+ miles. 841 fewer injury-causing crashes.

Our latest safety data shows the Waymo Driver continues to make roads safer for everyone.

Compared to human drivers across 5 territories, it reduced:
📉 Injury crashes by 82%
📉 Serious injury crashes by 95%

Full data:… pic.twitter.com/ojdZFptDdG

— Waymo (@Waymo) September 24, 2026

The company says that distance is equal to more than 300 human lifetimes of driving. The figures cover five cities: Atlanta, Austin, Los Angeles, Phoenix, and San Francisco.

What the Data Shows

Waymo compared its crash rates to human driving benchmarks in the same cities. The company says its vehicles had 82% fewer injury-causing crashes than human drivers.

Waymo also reported 95% fewer crashes involving serious injury or worse. That works out to about 20 times fewer severe crashes compared to human drivers.

The company said its vehicles helped avoid 841 injury-causing crashes where a human driver likely would have been involved in one. Waymo broke this down further by road user type.

Crashes involving pedestrians dropped 93% compared to human drivers. Crashes involving cyclists dropped 86%. Crashes involving motorcyclists dropped 82%.

Airbag deployments across all crash types were also down 82%. Some cities showed different results than others.

Atlanta saw 8% fewer crashes involving airbag deployments compared to human data. San Francisco saw 10% fewer injury crashes.

How Waymo Measured the Numbers

Waymo’s method compares its own crash rates to human crash data over the same distance traveled. This is meant to give a fair per-mile comparison.

The company pulled human crash numbers from state police reports and vehicle miles traveled data in the same operating areas. But Waymo noted a difference in how crashes get counted.

Self-driving companies like Waymo must report any physical contact that causes property damage. Human drivers are not always held to that same standard.

The National Highway Traffic Safety Administration estimates that 60% of property damage crashes involving human drivers go unreported. About 32% of injury crashes also go unreported.

This means some human crash numbers used for comparison may be lower than what actually happened. Waymo said aligning crash definitions between self-driving and human data is one of the hardest parts of building a fair comparison.

The company used prior safety research to help set consistent definitions across both data sets. Waymo said police report databases remain the most reliable source for human crash data.

Waymo acknowledged that its safety framework is not finished. The company said proving an “absence of unreasonable risk” is an ongoing part of its work, not a one-time result.

The Insurance Institute for Highway Safety weighed in on the release. IIHS chief research officer David Zuby praised Waymo for making detailed crash and mileage data public.

Zuby said this kind of openness supports independent research. He added that it also helps build public trust in the technology.

Zuby said he hopes other companies building self-driving systems will publish similar data. Waymo has not announced when it plans to release its next safety report.