Waymo’s driverless vehicles were involved in 68% fewer police-reportable crashes per mile than human drivers operating in the same cities and years, according to a new Insurance Institute for Highway Safety study. The analysis covered nearly 50 million driverless Waymo miles in Phoenix, San Francisco, Los Angeles and Austin from 2021 through 2024.
Waymo vehicles also had 81% fewer injury crashes and 85% fewer single-vehicle crashes per vehicle mile traveled. Their rate of rear-ending another vehicle was 91% lower, while their rate of being rear-ended was 40% lower. Results varied by market. Waymo’s overall crash rate was 76% lower in Phoenix, 35% lower in San Francisco and 71% lower in Los Angeles. The rate was slightly higher in Austin, where researchers said the available sample was small.
The findings could affect how auto insurers and claims professionals evaluate loss frequency, liability and injury exposure involving autonomous vehicles. Lower crash rates do not eliminate complex claims questions. Adjusters may still need to determine whether the automated driving system was active, whether remote assistance was involved, which vehicle initiated the collision and whether software, sensors, road conditions or another driver contributed to the loss.
Researchers faced major data problems when comparing autonomous vehicles with human drivers. Federal reports included duplicate records, incidents on private property, crashes that occurred while automation was disengaged and minor contacts that most motorists would not report to police. After reviewing and cleaning the data, the researchers determined that only 22% of 736 qualifying Level 4 crash involvements were likely or possibly police-reportable.
The study also found that autonomous vehicle operators are more likely to document minor impacts because of federal requirements and the high cost of sensors mounted on the vehicles. Human drivers frequently do not report comparable crashes, particularly when damage appears minor or when they want to avoid an insurance premium increase. These differences can distort crash-rate comparisons and create challenges for insurers using public data to evaluate autonomous vehicle losses.
Only Waymo provided enough mileage information for researchers to calculate a crash rate. Other companies reported crashes but did not release detailed vehicle-mile data. IIHS called for mandatory mileage reporting by location, month and operating mode, along with standardized fields for crash type, vehicle control, injury severity and property damage.
For claims organizations, consistent reporting would make it easier to compare autonomous vehicle performance, identify recurring loss patterns and evaluate whether a vehicle was the striking party or primary contributor. It could also support more reliable reserving, underwriting and subrogation decisions as robotaxi operations enter additional markets.
The study’s conclusions apply only to Waymo’s Level 4 vehicles operating without a human driver in the locations and conditions examined. Researchers cautioned that the results may not apply to other autonomous vehicle developers, new deployment areas or future operating conditions.
