Self-Driving Car Accident Statistics

Last reviewed: June 2026. Figures on this page come from the NHTSA Standing General Order (ADS and ADAS) crash dataset, which is updated periodically, and the named sources cited below. Cumulative totals continue to update, so confirm the latest numbers with each source before relying on them.

The technology behind self-driving cars continues to advance. And while a lot of road testing still needs to happen with autonomous vehicles, almost all the major U.S. and international automakers are heavily investing in self-driving tech. By all accounts, it won’t be long until driverless cars must merge with humans on the highway.

Self-driving cars have tremendous potential to make travel more convenient and improve road safety overall. But with new technology comes new concerns. What happens when things don’t work as intended? Are autonomous vehicles safer than human drivers?

At KNR, our car accident lawyers wanted to learn more about self-driving cars, their risks, and the future. Here is what we found about self-driving car accident statistics, how liability works, and what it means for injured drivers in Ohio.

The Status of Self-Driving Cars in the U.S.

It is hard to get precise figures, but there are currently only a few thousand autonomous vehicles on U.S. roads. Most of these are being used to gauge their self-driving capabilities.

As of 2021, 80 companies were testing approximately 1,400 self-driving cars, trucks, and other vehicles in 36 states. But in January 2023, Mercedes-Benz became the first automaker in the U.S. to receive government approval for a Level 3 driving feature in Nevada. Level 3 automation is significant because the car does all the driving, and a driver only needs to take control if necessary. This is a major advance from Level 2, which requires drivers to keep their eyes on the road.

But despite still being in the test phase, the market for fully automated cars is only increasing.

Every year, the autonomous vehicles industry increases by 16% globally, with the worldwide AV market presently worth $54 billion. According to PolicyAdvice, 33 million automated vehicles will be on the road by 2040.

The American public also seems to see the writing on the wall. Approximately 55% of Americans believe that most cars will be able to drive themselves by 2029 (AAA Annual Survey).

SAE Levels of Vehicle Automation

Driver-assist technologies are standard in most newer cars. These features keep us from drifting out of our lane, help us stop in time with automatic braking, alert us about a vehicle in our blind spots, and even discourage distracted driving practices. More advanced features can assist drivers with parking and offer adaptive cruise control. The Society of Automotive Engineers (SAE) groups these capabilities into six levels, from no automation to full automation.

Level 0 – Momentary Driver Assistance: Features are assistive and do not operate the vehicle. Drivers must steer, brake, and accelerate. Examples are:

• Automatic emergency braking
• Forward collision warning
• Lane departure warning

Level 1 – Driver Assistance: The driver is responsible for driving. When engaged, the system can perform either steering OR acceleration and braking. Examples are:

• Adaptive cruise control
• Lane-keeping assistance

Level 2 – Additional Assistance: The system can perform steering AND acceleration and braking when engaged. Examples are:

• Highway pilot

Level 3 – Conditional Automation: The system performs driving tasks while the driver remains available to take over.

Level 4 – High Automation: The system is fully responsible for driving while occupants act only as passengers and do not need to be engaged.

Level 5 – Full Automation: When engaged, the system handles all driving tasks while you, now the passenger, are not needed to maneuver the vehicle. The system can operate the car universally, under all conditions and on all roadways.

Most of the vehicles in self-driving car accidents today are Level 2 driver-assist systems, not fully autonomous cars, which matters when sorting out who was responsible.

Automated Vehicle Safety and Public Opinion

The AV industry is growing fast, but the safety of self-driving cars still appears to be under debate.

According to PolicyAdvice, 43% of Americans are uncomfortable inside a driverless car, citing safety as their biggest concern. A survey by Advocates for Highway and Auto Safety indicated 75% of people would rather drive themselves than ride in an AV. 71% said they would miss driving if autonomous driving became the norm.

Are Self-Driving Cars Safer Than Human Drivers?

Human drivers are still involved in the vast majority of crashes. NHTSA estimates that 39,345 people died in U.S. traffic crashes in 2024, down about 3.8% from 40,901 in 2023 (NHTSA, “NHTSA Estimates 39,345 Traffic Fatalities in 2024,” nhtsa.gov). Thousands more people are killed and injured in U.S. car accidents every year. We can do more to make roads safer, but is removing humans from the driver’s seat the correct answer?

Human behavior plays a major role. NHTSA’s National Motor Vehicle Crash Causation Survey (2015) found that a driver-related factor was the “critical reason,” the last event in the chain leading to a crash, in an estimated 94% of crashes. NHTSA cautions that this critical reason is not intended to be the cause of the crash or an assignment of fault, so the figure describes where the causal chain ended, not blame. Even so, it shows how often human factors are part of the picture, and why some hope automation can reduce crashes.

The Insurance Institute for Highway Safety looked at more than 5,000 car accident reports and found that while current self-driving technology could reduce accidents by around 34%, autonomous vehicles still have trouble avoiding, planning, deciding, and execution-related driving errors.

Reducing car accidents by over 30% is impressive, especially the impact AVs can have on drunk driving and distracted driving accidents. But the development of self-driving cars needs to account for road conditions and possible obstructions, Ohio traffic laws, and mechanical failures like a flat tire that could still cause an accident.

Self-Driving Car Accident Statistics

Since 2021, NHTSA’s Standing General Order has required carmakers and operators to report crashes involving automated driving systems (ADS) and SAE Level 2 driver-assist systems (ADAS). That dataset now holds more than 2,000 crash reports filed since 2021 (NHTSA Standing General Order). Because the order captures minor and no-injury events and updates continuously, the running total climbs steadily, and the exact cumulative count should be confirmed against the live tracker before it is cited.

One distinction matters when reading these numbers. The large majority of reported fatalities in the dataset involve SAE Level 2 driver-assist (ADAS) systems, which keep a human driver responsible for the car at all times. Fully self-driving (ADS) vehicles, where no human is driving, account for just 2 confirmed fatalities as of April 15, 2026, both associated with Waymo, which is also linked to 117 injuries (NHTSA Standing General Order data, via ConsumerShield).

Monthly reporting gives a clearer sense of the recent pace. Through 2025, monthly reported AV and ADAS crash counts ran roughly 61 to 112, reaching 110 in May 2025 and 112 in November 2025, then a record 118 in January 2026 (NHTSA Standing General Order, via ConsumerShield). These totals update continuously.

Source: NHTSA Standing General Order, via ConsumerShield trend tracker. Totals update continuously as new reports are filed and amended.

Early reporting was dominated by one manufacturer. In the first Standing General Order reporting window, July 1, 2021 to May 15, 2022, about 273 to 274 of 522 reported crashes, roughly 52%, involved Tesla vehicles (NHTSA Standing General Order). That snapshot reflects Tesla’s large SAE Level 2 ADAS fleet during that period, not a current figure. As driverless robotaxi mileage has grown, more recent reporting shows Waymo accounting for a larger share of fully driverless (ADS) incidents.

Source: NHTSA Standing General Order. This is the July 2021 to May 2022 snapshot and reflects that early, ADAS-heavy reporting window, not current company shares.

It is important to read this data with care. Reporting does not capture everything and can skew comparisons. There is often no way to confirm whether a self-driving or driver-assist feature was engaged at the moment of a collision, the data does not establish whether the other vehicle caused the crash, and many of these systems are still early in their deployment. In a large share of the no-injury ADS reports, the automated vehicle was rear-ended by another car. Because the Standing General Order dataset updates periodically, these totals change over time, so check the latest figures before relying on a specific number.