Tesla’s Robotaxi Safety Report: 3.1 million miles, read closely

On October 21, 2026, Tesla published the first edition of its Robotaxi Safety Report, which says its Robotaxi fleet drove 3.1 million miles with no safety monitor in the car between January 22 and October 18, 2026, with zero people injured, zero airbag deployments and zero pedestrians or cyclists hit (as reported by Tesla North and Not a Tesla App, both October 21). Austin drove about two thirds of those miles, and the report lists Las Vegas, Jacksonville, Phoenix and San Antonio as “Coming Soon” with no dates. These are Tesla’s own numbers, counted Tesla’s way. Below are the numbers, then the counting rules that matter when you read them.

What the report is

The report lives at tesla.com/robotaxi/safety. That page did not load for our fetcher, so every figure on this page comes from Tesla’s report as reported by Tesla North, Not a Tesla App and Drive Tesla Canada (all October 21, 2026), which agree on the numbers below.

• Window. Robotaxi rides from January 22 to October 18, 2026.
• What counts as a mile. Only miles driven with no safety monitor inside the car, including trips where a remote operator was assisting or had taken over (Tesla North).
• Which cars. Both the Model Y robotaxis and Cybercabs (Not a Tesla App).
• Which collisions. Every collision Tesla filed with NHTSA under its Standing General Order on crash reporting while a Robotaxi drove with no safety monitor (Tesla North).
• Live counter. The page also shows a counter that estimates miles driven since October 18. Not a Tesla App reported it had already passed 3.3 million when it wrote its story. The 3.1 million figure is the one tied to the report’s data.

Miles by city

Tesla broke the 3.1 million miles out by city. The share column is our arithmetic, not Tesla’s: each city’s miles divided by the six-city sum of 3,112,541, rounded to one decimal.

Tesla North also reports that the report breaks out 862,000 miles driven after dark, 113,000 miles in rain and 575,000 miles in weekday rush hour. The San Francisco Bay Area is not in the table: Tesla runs rides there with a safety driver behind the wheel, so those miles are not driverless and the report leaves them out (Not a Tesla App). For how the fleet count behind these miles has moved, see our Tesla robotaxi count ledger.

The safety claims, next to Tesla’s benchmark

Tesla compares its Robotaxi results with what it says human drivers would have done over the same miles, using Texas police crash data (Tesla North).

The two columns are not counted the same way, which is the first thing the next section covers.

How to read it: six counting rules

1. Tesla’s own numbers. The report is Tesla’s count of its own fleet. None of the coverage we read describes an independent audit or third-party verification; Drive Tesla Canada noted the claims “will remain subject to independent scrutiny as the fleet expands.”
2. The injury count is narrower than the benchmark. Tesla’s injury metric counts only collisions where a Robotaxi made first contact, while the human benchmark includes collisions regardless of which vehicle made contact (Drive Tesla Canada). A crash where another car hits a Robotaxi and someone is hurt can sit in the benchmark’s kind of number but not in Tesla’s.
3. Remote-assisted trips are in. Miles where a remote operator was assisting or had taken over still count as autonomous miles, and collisions during them are in the data (Tesla North; Drive Tesla Canada). We explain what remote help looks like in what happens when a Cybercab gets stuck.
4. Monitor-era miles are out. The report leaves out crashes and miles from rides that had a safety monitor on board, most of them before January 22 (Tesla North).
5. The Bay Area is out. Rides there have a safety driver, so they are excluded (Not a Tesla App).
6. Two events were excluded from injuries. One where another vehicle pushed an SUV into a Robotaxi stopped at a stop sign, sending an occupant of that other vehicle to the hospital; and one where a rider claimed injury after the car drove over a pothole, which Tesla says its investigation did not support (Tesla North).

None of this makes the zeros false or true. It tells you what the zeros measure: injuries in collisions a driverless Robotaxi started, in Texas and Florida, from January 22 to October 18, 2026, by Tesla’s count.

The “Coming Soon” cities

The report lists four cities as “Coming Soon”: Las Vegas, Jacksonville, Phoenix and San Antonio. Tesla gave no launch date for any of them (Not a Tesla App). Las Vegas and Phoenix would add Nevada and Arizona to a driverless service that so far runs only in Texas and Florida (Tesla North). Our Las Vegas post covers what Nevada has approved so far.

What regulators have open

Two NHTSA files sit alongside the report. Audit Query AQ26002, opened September 3, 2026, reviews “the process and technical data on which Tesla relied when certifying the Cybercab,” an estimated 1,000 vehicles; an audit query is not a recall or a defect finding, and Tesla says it certified the Cybercab as meeting all applicable federal safety standards (NHTSA; Electrek). Defect petition DP26010, opened October 9, 2026, alleges Tesla Robotaxis exceed posted limits in active school zones; NHTSA lists four incidents and no crashes, injuries or deaths, and has not decided whether to open a full investigation (Tesla North; Insurance Journal). The full background is in our Cybercab NHTSA investigation explainer.

What it means for owners

Florida, where our lead market is, supplied 160,432 of the 3.1 million miles, about 5% (our arithmetic from Tesla’s city figures). Orlando alone was about 2%. Most of what the report measures happened in Austin.

The report is about Tesla’s own fleet. It says nothing about privately owned cars, owner revenue, or how many hours a day each car carried riders. On Tesla’s October 21 earnings call, the only owner-car timing given was a spoken target of “somewhere around Q2 next year” for the first customer vehicles, with no published terms; our Q3 2026 earnings write-up has the quotes.

And a car that drives itself 3.1 million miles still has to stop somewhere. Every one of those robotaxis needs a place to park, charge and get cleaned between rides, and the report does not say who would do that for a car an owner supplies.

DockDuty sells depot stalls to people planning to own robotaxis, so it is not a neutral party; nothing here is investment advice.

DockDuty status

Orlando is our lead market, with its depot targeted for Q4 2026. We're planning locations in several more metros at the same time, including depots in Miami and Tampa Bay. None has a site, lease or date yet.

No DockDuty depot is operating and no customer vehicle is earning. A stall reservation is $450, fully refundable on request any time before onboarding. Reserve a stall.

Tesla Robotaxi Safety Report: FAQ

What is Tesla’s Robotaxi Safety Report?

A page Tesla published on October 21, 2026, at tesla.com/robotaxi/safety. Its first edition covers Robotaxi rides from January 22 to October 18, 2026, with no safety monitor in the car, and gives miles by city, collision results, a human-driver comparison and a list of “Coming Soon” cities, as reported by Tesla North and Not a Tesla App.

How many miles have Tesla robotaxis driven?

3.1 million autonomous miles as of October 18, 2026, per Tesla’s report: Austin 2,118,015, Dallas 608,130, Houston 225,964, Orlando 61,053, Tampa 53,117 and Miami 46,262. The count includes trips where a remote operator assisted or took over, and excludes miles with a safety monitor on board and the Bay Area, where a safety driver is required.

Has a Tesla robotaxi injured anyone?

Tesla’s report says zero people were injured by a Robotaxi from January 22 to October 18, 2026, against about 9.1 injury collisions it says human drivers would have had over the same miles. Its injury metric counts only collisions where the Robotaxi made first contact, while the human benchmark counts all, and Tesla excluded two events: an SUV pushed into a stopped Robotaxi by another vehicle, and a pothole injury claim it says its investigation did not support.

Is Tesla’s robotaxi safety data independently verified?

None of the coverage we read describes an independent audit. The figures are Tesla’s own, built from the collisions it filed with NHTSA under the Standing General Order on crash reporting.

What are Tesla Robotaxi’s next cities?

The report lists Las Vegas, Jacksonville, Phoenix and San Antonio as “Coming Soon,” with no launch date for any of them.

Does the report include privately owned cars?

No. It covers Tesla’s own Robotaxi fleet, Model Y robotaxis and Cybercabs, and says nothing about owner cars, owner revenue or how much each car is used.