Waymo’s self-driving safety data reveals 82% fewer injury crashes than humans

Waymo just handed the public a mountain of numbers, and they’re hard to ignore. The company says its self-driving vehicles have now logged 270 million fully autonomous miles through June 2026, and according to its own safety comparisons, those miles came with dramatically fewer injury crashes than what human drivers rack up over the same distance.

Key takeaways

• Waymo released safety data covering 270 million fully autonomous miles through June 2026, spanning Atlanta, Austin, Los Angeles, Phoenix, and San Francisco.
• The company reports 82% fewer injury-causing crashes and 95% fewer serious injury crashes compared with human drivers over the same distance.
• According to Waymo, incidents that would likely have led a human driver to be in one of 841 injury-causing crashes were successfully avoided by its vehicles.
• Pedestrian crashes dropped 93%, cyclist crashes dropped 86%, and motorcyclist crashes dropped 82% versus human driving benchmarks.
• The Insurance Institute for Highway Safety praised Waymo for publishing the detailed data openly.

Waymo Releases Extensive Autonomous Driving Safety Data

Waymo’s latest release is one of the largest safety data drops the company has published to date. The company announced the figures on social media, framing them as proof that its driving system keeps improving road outcomes for everyone, not just its own passengers.

Scope of Data and Geographic Coverage

The dataset covers five U.S. cities: Atlanta, Austin, Los Angeles, Phoenix, and San Francisco. Waymo says the 270 million autonomous miles behind the numbers equal more than 300 human lifetimes of driving, giving the comparison a scale that’s hard to replicate with smaller pilot programs.

Key Safety Performance Metrics

Waymo’s headline figures are stark. The company reports 82% fewer injury-causing crashes than human drivers over the same distance, and 95% fewer crashes involving serious injury or worse — roughly 20 times fewer severe crashes compared to human benchmarks in the same cities.

When analyzed by miles driven, Waymo’s data suggests its vehicles prevented 841 crashes involving injuries that a human motorist would probably have experienced. The company also split results by road user type: pedestrian-related crashes fell 93%, cyclist crashes dropped 86%, and motorcyclist crashes fell 82% compared with human drivers.

Airbag deployments, which serve as a rough proxy for crash severity, were down 82% overall. That figure wasn’t uniform across every city, though. Atlanta saw 8% fewer crashes involving airbag deployments compared with human data, while San Francisco recorded 10% fewer injury crashes — a reminder that autonomous vehicle crash data can vary meaningfully depending on local traffic patterns, road design, and driving culture.

Methodology and Challenges in Data Comparison

Comparing robot drivers to human drivers sounds simple, but Waymo’s own numbers show why it’s anything but. The company built its comparisons using state police reports and vehicle miles traveled data from the same operating areas, aiming for a fair per-mile benchmark rather than a raw crash count.

Here’s where things get complicated. Self-driving operators like Waymo are required to report virtually any physical contact that causes property damage, no matter how minor. Human drivers face no such blanket requirement, and a lot of fender-benders and minor injury crashes simply never make it into official records.

According to estimates from the National Highway Traffic Safety Administration, roughly 60% of crashes causing only property damage and around 32% of those causing injuries are never reported when human drivers are involved. That gap matters: it means the human crash baseline Waymo is measuring against may already understate how often people actually crash, which would make Waymo’s relative improvement look smaller than it really is in practice.

Waymo admitted that reconciling how crashes are defined across autonomous and human-driven datasets ranks among the toughest challenges in creating an equitable comparison, so the company drew on existing safety studies to establish uniform definitions for both. The company also said police report databases remain the most reliable source available for human crash figures, even with their known gaps. Notably, Waymo has stopped short of calling its safety framework finished — the company described proving an “absence of unreasonable risk” as ongoing work rather than a box it has already checked.

Industry Recognition and Implications for Public Trust

Independent validation is where this release gets its real weight. The Insurance Institute for Highway Safety weighed in on the disclosure, and its endorsement is likely to matter more to skeptical regulators and riders than anything Waymo says about itself.

IIHS chief research officer David Zuby praised Waymo for making its detailed crash and mileage data public. He explained that this level of transparency not only enables independent researchers to conduct their own analyses but also fosters greater public confidence in the technology, and Zuby expressed hope that competitors developing similar self-driving systems would follow suit by releasing comparable data.

That last point is worth sitting with. In an industry where companies routinely guard operational data as a competitive edge, Waymo’s willingness to publish granular crash and mileage figures — including the messy parts, like city-by-city variability and unresolved reporting gaps — sets a bar that rivals in the self-driving space haven’t matched with comparable transparency. Whether that becomes an industry norm or stays a Waymo-specific move remains to be seen; the company hasn’t said when its next safety report will land.

FAQ

How much safer are Waymo’s self-driving cars compared to human drivers?

Waymo’s vehicles had 82% fewer injury-causing crashes and 95% fewer serious injury crashes compared to human drivers over the same distance.

What geographic areas does Waymo’s safety data cover?

The safety data covers five U.S. cities: Atlanta, Austin, Los Angeles, Phoenix, and San Francisco.

How does Waymo compare its crash data to human driver crash data?

Waymo compares its crash rates using state police reports and vehicle miles traveled data, aiming for a fair per-mile comparison with human driving benchmarks.

What has been said about the transparency of Waymo’s data release?

The Insurance Institute for Highway Safety praised Waymo for publishing detailed crash and mileage data openly, which supports independent research and helps build public trust.

Article produced with the assistance of artificial intelligence and reviewed by the editorial team.