Self-Driving Car Accident Statistics: 2026 Report

Robotics Reviewed Research DeskNovember 16, 2026Sources: NHTSA, Waymo, Baidu

Our analysts compiled self-driving car accident statistics from crash reports that companies must file with the National Highway Traffic Safety Administration (NHTSA), from safety data published by Waymo and Baidu, and from NHTSA traffic death estimates. Self-driving here means an automated driving system (ADS), not driver-assist features such as Tesla Autopilot.2,3

Key Findings

  • Self-driving companies filed 1,551 crash reports with NHTSA between June 16, 2025 and September 15, 2026.2,3
  • 83% of those reports involved property damage with no injury reported, and 2 alleged a fatality.2
  • In 51% of the reports, the self-driving vehicle was stopped just before the crash.2
  • Waymo had 82% fewer injury crashes and 95% fewer serious injury crashes than human drivers over the same 271.3 million miles, through June 2026.4
  • Baidu says its fully driverless Apollo Go fleet averaged one airbag deployment per 14.4 million kilometers as of June 2026.5

In the following sections, we cover NHTSA crash reports, Waymo crash rates, and human driver benchmarks.

Self-Driving Crash Reports Filed With NHTSA

Under NHTSA's Standing General Order, companies must report a crash if an automated driving system was in use within 30 seconds of it and the crash caused certain property damage or injury. A report does not mean the self-driving car caused the crash.3

Self-driving crash reports filed with NHTSA by company, June 2025 to September 2026
Waymo
1,283
Avride
81
Zoox
78
Tesla
36
May Mobility
20
Motional
15

Source: NHTSA Standing General Order ADS incident report data, latest version of each report. Counts are not adjusted for miles driven. Chart: Robotics Reviewed.

Key Findings

  • Waymo filed 1,283 of the 1,551 reports, or 83%. Waymo also drives far more driverless miles than any other company.2,4
  • Avride filed 81, Zoox 78, Tesla 36, May Mobility 20, and Motional 15.2
  • California accounted for 808 reports, Texas 322, and Arizona 208.2
  • 157 reports alleged an injury or death, including 4 serious injuries with hospitalization and 2 fatalities.2
  • Both fatal reports came from Waymo. In Dallas in August 2026, an SUV struck a pedestrian who was thrown toward a slowing Waymo, and in Tempe in September 2025, a motorcycle approached a slowing Waymo from behind, according to Waymo's report narratives.2
  • 26 reports involved a pedestrian, cyclist, scooter or skateboard rider, or other non-motorist.2
Highest injury severity alleged in self-driving crash reports, June 2025 to September 2026
SeverityReportsShare
Property damage, no injury reported1,29583.5%
Minor injury, no hospitalization996.4%
Minor injury, hospitalization392.5%
Moderate injury130.8%
Serious injury, hospitalization40.3%
No injury reported (other)744.8%
Unknown251.6%
Fatality20.1%

Source: NHTSA Standing General Order ADS incident report data. Table: Robotics Reviewed.2

Waymo Crash Rates Compared With Human Drivers

Waymo publishes crash rates for its driverless miles and compares them with human crash rates on the same kinds of roads in the same cities. The comparison counts every crash, regardless of who was at fault.4

Injury crashes per million miles, Waymo and human driver benchmark, through June 2026
0.002.004.006.008.00All Waymo cities3.77San Francisco6.71Atlanta area6.44Austin3.25Los Angeles2.49Phoenix1.95
  • Waymo
  • Human driver benchmark

Source: Waymo Safety Impact hub (rider-only miles; human benchmarks for the same roads). Chart: Robotics Reviewed.

Key Findings

  • Across all its cities, Waymo had 0.67 injury crashes per million miles, against 3.77 for human drivers.4
  • Waymo had 93% fewer pedestrian injury crashes, 86% fewer cyclist injury crashes, and 82% fewer motorcycle injury crashes than human drivers.4
  • Waymo had 82% fewer crashes with an airbag deployment in any vehicle, or 357 fewer over the same distance.4
  • In 41% of Waymo crashes reported to NHTSA, the change in speed on impact was less than 1 mile per hour.4
Waymo crash reductions against human drivers, through June 2026
Crash typeFewer than human driversCrashes avoided
Serious injury or worse95%55
Airbag deployment in any vehicle82%357
Any injury reported82%841
Pedestrian injury93%87
Cyclist injury86%56
Motorcycle injury82%36

Source: Waymo Safety Impact hub. Table: Robotics Reviewed.4

Human Driver Crash Benchmarks

Self-driving crash counts are small next to crashes caused by human drivers in the United States.6

Key Findings

  • An estimated 36,640 people died in US traffic crashes in 2025, 6.7% fewer than in 2024.6
  • The 2025 US fatality rate was 1.10 deaths per 100 million vehicle miles traveled, the second lowest on record.6
  • 39,254 people died in US traffic crashes in 2024, according to NHTSA's final annual data.6
  • Human drivers in Waymo's cities had 0.21 serious injury crashes per million miles, against 0.01 for Waymo.4
Sources
Methodology
Crash report counts use the latest version of each report in NHTSA's ADS incident report file for the third amended Standing General Order, which covers reports from June 16, 2025 to September 15, 2026 (1,551 unique reports). Reports are filed by the operating company, can include crashes caused by other drivers, and are not adjusted for miles driven, so counts by company reflect fleet size as much as safety. Severity is the highest injury severity alleged in the report. Waymo crash rates and reductions are Waymo's own figures for rider-only miles through June 2026, compared with human benchmarks for the same areas. Apollo Go figures are as stated by Baidu.