Data shared by Tesla regarding its self-driving robotaxis shows that the automaker’s autonomous vehicles are involved in almost four times more accidents on average than human drivers. This is according to the Electrek outlet, which analyzed the data combined with that provided by the NHTSA.
This performance is in sharp contrast with those of other self-driving-vehicle operators like Waymo, which have reported collision rates below those recorded by human drivers. (See below for a comparison between the systems in use by Tesla and Waymo).
Here’s how the data led to the outlet’s conclusion: Tesla reported in its final earnings report of 2025 that its Tesla Robotaxi service, currently operating in limited form in Austin, Texas, accumulated a total of 500,000 miles (804,700 km) driven between June and November of 2025. Separately, the NHTSA (National Highway Traffic Safety Administration) received reports of nine collisions involving robotaxis from the EV maker.
That works out to one accident per 55,000 miles (89,000 km). NHTSA data puts the accident rate involving human drivers at one accident per 500,000 miles (804,700 km). That last figure does not include human-involved accidents that aren’t reported to authorities, however. Electrek’s adjustment to account for that leaves an accident rate of roughly one accident per 200,000 miles (321,900 km) driven.
And that makes Tesla’s robotaxis four times as likely to be involved in road accidents than vehicles driven by humans.
It's worth mentioning as well that the performance of Waymo’s driverless vehicles, accumulated across some 125 millions of miles driven, has been without the presence of human safety monitors, who are still present in Tesla’s robotaxis and have presumably been able to prevent some accidents, lowering the reported rate for that company’s fleet.
Electrek does point to some potentially encouraging news regarding the safety performance of Tesla’s robotaxis: of the nine incidents reported by the EV maker, only two came in the last two months of the period reported – which could be interpreted to mean there’s progress being made. It’s not much, but it’s something…
Full speed ahead
Tesla won’t be limited for long to the Austin area. The company has plans to expand its robotaxi service in the first half of 2026 to other major urban centres like Phoenix, Las Vegas, Miami and Orlando in Florida and Dallas and Houston in Texas.
The company is also advancing with plans to begin production of the Cybercab, designed specifically as a self-driving taxi, at its plant in Texas.
How the Tesla and Waymo self-driving systems differ
Tesla: Betting on "Vision Only"
Tesla's self-driving system relies exclusively on computer vision (high-definition cameras) and artificial intelligence (neural networks). This means no radar or Lidar, both of which Elon Musk early on deemed too expensive and unnecessary.
Critics of the system point out that cameras can be blinded by the sun or obstructed by fog or snow. They can also sometimes struggle to evaluate millimetre-accurate distances or unusual shapes (such as a white truck against a bright sky).
Without backup sensors — and thus a lack of redundant tools — the AI must "guess" depth, which increases the margin of error.
Waymo, for its part, uses a comprehensive sensor suite that includes Lidar, radar and cameras. This hardware redundancy allows the system to validate every decision through three different sources, and it explains why Waymo can operate without a safety driver on board (so, Level 4), whereas Tesla still requires one.