Report: Waymo Driverless Cars Record Fewer Crashes Than Human-Driven Cars

Have driverless cars lived up to their promise? More cities are weighing whether it’s safe to open their streets to Waymo or other autonomous vehicle companies. Driverless car crashes and mishaps draw widespread attention, and just the idea of a car with no human driver makes some people wary. But a new study from the Insurance Institute for Highway Safety finds that Waymo’s driverless cars’ track record so far looks pretty good.

Waymo has the biggest autonomous vehicle fleet in the U.S. It currently has vehicles operating in 11 cities, primarily in Arizona California and Texas, and it’s eyeing expansion to another 20 U.S. cities as well as cities in Japan and the U.K.

Compared to human-driven cars, the company’s cars — when operating in driverless mode — were less likely to be involved in collisions significant enough to merit reporting to the police, the report said. When involved in such collisions, Waymo cars were the primary cause of the crash only 8 percent of the time; in another 8 percent of incidents it was unclear who was primarily to blame.

To understand how likely crashes were, the study considered how many crashes cars got into per mile they traveled. Researchers focused on Waymo over other autonomous vehicle companies because it was the only one to publish data on vehicle miles traveled for its fleet. The study compared human and Waymo crashes from 2021-2024 in areas of Arizona, California and Texas where the company operated.

Key Findings

• Compared to human-driven cars operating in the same areas, driverless Waymo cars had a 68 percent lower “crash rate.”
  
  • That rate varied by city: In Phoenix, Waymo had a 76 percent lower crash rate than human drivers; 71 percent lower in Los Angeles; 35 percent lower in San Francisco; and 3 percent higher in Austin. (Researchers caution that they had a small sample size for both Waymo and human drivers in Austin, which could have affected results).  
• Waymo cars operating in driverless mode also got into fewer collisions severe enough to merit reporting to the police. From 2021-2024, Waymo driverless cars were involved in an average of 1.28 such crashes per million vehicle miles traveled, while human-driven cars were involved in 4.06.
• Compared to human-driven cars, Waymo cars were involved in 85 percent fewer single-vehicle crashes per vehicle miles traveled.
• Waymo cars were also involved in 81 percent fewer injury- or fatality-causing crashes per vehicle miles traveled, compared to human-driven cars.

Researchers note the study doesn’t compare how a driverless car performs against an ideal human driver, but against the whole range of people who get behind the steering wheel, including those who are aggressive, distracted or impaired. And the lower injury rates for crashes involving autonomous vehicles could partially reflect that they may not have been carrying any passengers or drivers inside to be injured.

Waymo cars didn’t drive on interstates or freeways during 2021-2024, so the study only compared crash rates on other kinds of roads. It can’t say how the autonomous vehicles would fare on interstates, highways or in different cities — but that kind of data will be important to collect.

Getting such data could require rethinking crash reporting, study authors say. Inconsistencies in how collision data is reported for crashes involving human-operated cars versus driverless ones required researchers to take a time-consuming approach to analyzing autonomous vehicle-involved collisions. That approach could be difficult to replicate in future studies that would necessarily involve more driverless car crash data, as such vehicles become more prevalent.