The short version
- Waymo claims an 82 percent reduction in injury-causing crashes based on 271 million driverless miles across five cities.
- Independent analysis by the Insurance Institute for Highway Safety supports lower crash rates but criticizes the adequacy of current data collection methods.
- Critics argue that self-reported metrics obscure traffic disruptions and fail to account for the advantage of advanced sensors over average human-driven vehicles.
Waymo has released a new set of safety statistics asserting that its autonomous vehicles are significantly safer than human operators. The company examined 271 million miles driven without human intervention across five major metropolitan areas through June 2026. This dataset represents an increase of 50 million miles from the previous reporting period. Waymo states that this volume is equivalent to more than 300 human lifetimes of driving, assuming an average lifetime mileage of 850,000 miles per person.
According to the company’s analysis, its robotaxis experienced an 82 percent reduction in crashes causing injuries compared to human drivers. The data indicates even steeper declines in severe outcomes, with a 95 percent drop in incidents resulting in serious injury or worse. Specific vulnerable road users also saw reduced risk, including a 93 percent decrease in pedestrian injury crashes and an 86 percent reduction in cyclist injury crashes. Motorcycle crash injuries involving Waymo vehicles fell by 82 percent relative to human-driven counterparts.
Despite these figures, the total mileage logged by Waymo remains a small fraction of national driving activity. Human drivers in the United States cover approximately 3 trillion miles annually. This disparity creates a statistical gap that complicates direct comparisons between autonomous systems and general traffic safety. While Waymo operates nearly 4,000 vehicles across more than ten cities, logging roughly two million miles weekly, the sample size is still limited when viewed against the broader transportation network.
Independent researchers have offered partial validation of Waymo’s claims. A study by the Insurance Institute for Highway Safety found that Waymo vehicles were involved in 68 percent fewer crashes per vehicle mile traveled than human drivers. However, David Harkey, president of the IIHS, cautioned that existing data collection frameworks are inadequate for monitoring large-scale deployment. He emphasized that current systems do not provide the continuous oversight necessary for expanding autonomous operations safely.
Experts from institutions such as Johns Hopkins University highlight structural issues in how safety data is reported across the industry. The lack of uniform guidelines makes it difficult to compare metrics between different companies or against human driving standards. Critics argue that comparing high-end, sensor-equipped robotaxis with average human-driven vehicles—which may include older models or distracted operators—primarily demonstrates the effectiveness of advanced braking and sensing technology rather than proving the superiority of autonomous decision-making software.
Phil Koopman, a safety expert at Carnegie Mellon University, suggests that aggregate safety dashboards can function as public relations tools. He argues that such summaries often downplay real-world complexities, including minor incidents, embarrassing traffic jams, and edge cases that do not result in physical contact but still disrupt traffic flow. These non-crash disruptions are not captured in standard crash statistics but contribute to the overall impact of autonomous vehicles on urban mobility.
Federal regulations require Waymo to report crashes defined as physical contact with other vehicles, people, or property. Under a 2021 standing general order, data from March to September 2026 shows that Waymo vehicles were involved in 797 such incidents across all operating cities. The company reported one fatality during this period, involving a pedestrian in Dallas who was struck by a human driver, hit the pavement, and then came into contact with a Waymo vehicle.
The release of this data comes as policymakers continue to debate the safety and economic implications of autonomous vehicles. While Waymo hopes these figures will persuade skeptical officials in cities like New York, Boston, and Washington, D.C., many remain hesitant. Concerns about traffic congestion and potential job losses persist in municipalities that have not yet approved full-scale operations. The company plans to update its online safety hub with the new driving figures, aiming to build support for broader adoption despite ongoing criticism regarding data transparency and completeness.
Sources behind this briefing
Go to the original reporting
- The Verge↗Waymo’s driverless cars continue to crash less often than people
