Tesla Reports Zero Robotaxi Injuries Across 3.1 Million Miles Under Narrower Rules

Tesla says its Robotaxi fleet completed 3,112,541 autonomous miles without causing an injury between January 22 and October 18, 2026. The safety report published Wednesday also records no Robotaxi airbag deployments and no collisions with pedestrians or cyclists but the company’s comparison with human drivers uses broader collision criteria for people than for its autonomous vehicles.

The difference sits inside the injury measure. Tesla counts a Robotaxi injury collision only when its vehicle made first contact. Its human-driver benchmark, derived from Texas police data, includes injury collisions regardless of which vehicle made first contact. The results therefore describe a notable period without an injury attributed under Tesla’s method, rather than a like-for-like demonstration that its Robotaxis are safer than human drivers.

What the 3.1 million miles cover

The mileage is substantial enough to provide useful operating data across more than ideal daylight conditions. Tesla reports 862,000 miles after dark, at least 113,000 miles in rain and 575,000 miles during rush hour. Austin contributed more than 2.1 million miles, while Dallas accounted for about 608,000.

These exposure categories matter because an autonomous-driving system depends on a chain of sensing, scene interpretation, trajectory planning and vehicle actuation. Darkness and rain change the visual information available to the system, while rush-hour traffic increases the density and interaction rate of surrounding road users. Mileage in those conditions is more informative than an undifferentiated total, although Tesla has not provided enough detail here to calculate separate safety rates for each condition.

The report page also carries a higher live mileage counter. Tesla describes that figure as an estimate based on the fleet’s current operating pace, so the fixed total of 3,112,541 miles is the defined reporting period rather than a continuously audited count.

The human comparison is not equivalent

For approximately 2.95 million miles in Austin, Dallas and Houston, Tesla estimates that human drivers would have experienced about 9.1 injury collisions, 4.1 collisions involving airbag deployment and 0.52 collisions with pedestrians or cyclists. Tesla says Texas police reports provide context for its results.

Those estimates cannot be placed directly beside Tesla’s zeroes without accounting for the different inclusion rules. The Robotaxi injury metric requires the autonomous vehicle to make first contact, while the police-derived human benchmark includes collisions in which another road user initiated contact. The comparison also excludes miles driven with onboard safety monitors and covers only Tesla’s Texas markets.

Zero injuries also does not mean zero collisions. Tesla says its reportable-event data include every collision submitted to the National Highway Traffic Safety Administration while a Robotaxi is operating without an onboard safety monitor, including events in which remote-operator assistance is being used. The company offers downloadable event-level data, allowing the collision count and circumstances to be examined separately from the headline injury measure.

Remote assistance is an important part of the system boundary. It can help resolve unusual situations without placing an onboard driver in the vehicle, but it also means fleet performance depends on communications, supervision procedures and operator availability as well as the vehicle’s onboard sensing and control stack. A useful safety assessment needs to track how often assistance occurs, what prompts it and whether the intervention changes the outcome.

Mileage alone cannot settle rare-event safety

Injury and fatal collisions are uncommon relative to total road mileage, making strong statistical claims difficult even when a fleet accumulates millions of miles. A widely cited study of autonomous-vehicle reliability calculated that hundreds of millions and in some cases hundreds of billions of miles could be required to demonstrate reliability for injuries or fatalities through road testing alone.

That does not make Tesla’s reported run irrelevant. It makes consistent definitions, event-level disclosure and operating-condition data essential. Fleet safety must be assessed through multiple layers: real-world outcomes, simulations, controlled tests, near-event analysis, remote-assistance records and performance in difficult but repeatable scenarios.

Tesla began its Austin Robotaxi service in June 2025 with onboard safety monitors and now lists operations in Austin, Dallas, Houston, Miami, Orlando and Tampa. Las Vegas, Jacksonville, Phoenix and San Antonio are identified as upcoming markets, without launch dates. Federal regulators continue to scrutinize the company’s Robotaxi operations as the fleet expands.

The company has also released first-responder resources covering training, round-the-clock Robotaxi support and windshield QR codes containing vehicle information. Texas already maintains a broader automated-vehicle recognition guide for first responders, reflecting a deployment constraint that mileage statistics cannot capture: a driverless vehicle must remain understandable and manageable when police, firefighters or paramedics take control of an incident scene.

Tesla’s 3.1 million injury-free miles are its clearest Robotaxi safety disclosure so far. The next meaningful step is a comparison built on identical collision definitions, alongside enough event and intervention data to show how the autonomous system and the human support behind it performs as exposure grows.

By Jonathan Barrett — Editor for AMI’s future mobility and autonomous systems section, with two decades covering robotics, e-mobility, drone-vehicle convergence, and transport mechanical systems.