Waymo's latest self-reported safety data, covering 271 million driverless miles through the end of June across Atlanta, Austin, Phoenix, Los Angeles, and San Francisco, claims an 82% reduction in crashes causing injury and a 95% reduction in crashes causing serious injury or worse, compared with human drivers in the same cities3,4,6.
The dataset adds 50 million miles to Waymo's previous end-of-March report. Waymo says the data translates to 841 fewer injury-causing crashes than human drivers would have produced over the same mileage2. Of those avoided crashes, 175 would have involved pedestrians, cyclists, or motorcyclists, and 55 would have caused serious injuries or worse.
Vulnerable road users
Breaking the figures down by road-user category, Waymo reports 93% fewer pedestrian crashes causing injury, 86% fewer cyclist crashes causing injury, and 82% fewer crashes involving motorcyclists. Waymo's peer-reviewed safety impact data reports roughly 94% fewer serious or fatal injury crashes versus the same-city human benchmark.
An independent study by the Insurance Institute for Highway Safety found Waymo vehicles were involved in 68% fewer police-reportable crashes per mile than human drivers across San Francisco, Phoenix, Los Angeles, and Austin. Waymo is now conducting roughly half a million rides per week and has logged over 20 million paid rides.
Limits of the dataset
The data remains self-reported, and some experts say it does not capture non-crash traffic disruptions caused by Waymo's robotaxis. Henry Kowal, director of insurance product at Arity, said the early data is "encouraging" but "limited"5. Waymo's service has only recently expanded into its first real winter-weather market, Denver, where its vehicles are currently designed to pull over rather than drive through snow. A number of Waymo vehicles became stranded in San Francisco this past July 4th amid holiday fireworks, heavy congestion, and a cell service blackout.
The IIHS said its monitoring system is not yet sufficient to track robotaxis at a full-scale rollout. Electrek noted that while Waymo's millions of miles seem substantial, human drivers cover roughly 3 trillion miles per year in the US, leaving a significant gap in statistical power.
ANALYSIS The 271-million-mile dataset is the largest Waymo has published, but the gap between that figure and the trillions of human-driven miles means the per-category crash reductions, particularly the 95% serious-injury figure, rest on relatively small absolute numbers of incidents. The Denver expansion and the July 4th stranding episode illustrate operational boundaries that the aggregate safety statistics do not address.
Tesla is currently running only a few dozen robotaxi vehicles compared with Waymo's roughly 4,000. Waymo resumed highway-speed autonomous rides in Austin after suspending the capability earlier this year[1].