How Many Self-Driving Cars Have Crashed? 2026 Robotaxi Safety Data

The Real Crash Numbers Behind Robotaxis

The question "how many self-driving cars have crashed" has no single, simple answer — but the data that does exist tells a nuanced story. Since the National Highway Traffic Safety Administration (NHTSA) began mandating crash reporting for vehicles equipped with Level 2 and higher automated driving systems in June 2021, thousands of incident reports have been filed. However, the vast majority of those reports involve Level 2 driver-assistance systems (like Tesla Autopilot and GM Super Cruise), not the fully driverless Level 4 robotaxis operated by companies like Waymo and Cruise.

Understanding this distinction is essential before examining any crash data. A Tesla Model Y on Autopilot with a driver behind the wheel is fundamentally different from a Waymo Jaguar I-PACE navigating Phoenix with no one in the driver's seat. Yet public crash databases often lump them together, creating confusion about what the numbers actually mean.

What NHTSA Data Actually Shows About Automated Vehicle Crashes

Under NHTSA's Standing General Order, manufacturers and operators must report crashes involving vehicles with SAE Level 2 through Level 5 automation. Since reporting began, NHTSA has collected data on over 5,000 incidents. But context matters enormously:

• ~90% of reported incidents involve Level 2 systems — vehicles like Teslas on Autopilot where a human driver is supposed to be supervising. These are not "self-driving car crashes" in the robotaxi sense.
• Level 4 robotaxi incidents are far fewer in absolute numbers, but these vehicles also drive far fewer total miles than the millions of Level 2-equipped consumer vehicles on the road.
• Many Level 2 reports involve rear-end collisions where the human driver was inattentive, not a system failure per se.

The reporting requirement itself inflates the numbers: even minor fender-benders at low speed must be reported if an automated system was engaged, whereas comparable human-driver crashes often go unreported to any central database.

Where Can You Actually Use a Robotaxi Right Now?

Fully driverless, commercial robotaxi service remains limited to a handful of U.S. metro areas. Here is the current landscape as of early 2026:

Robotaxi Service Availability — 2026

Operator	City/Region	Service Type	Hours	Status
Waymo One	Phoenix, AZ (metro-wide)	Fully driverless commercial	24/7	Live
Waymo One	San Francisco, CA	Fully driverless commercial	24/7	Live
Waymo One	Los Angeles, CA	Fully driverless commercial	Expanded hours	Live
Waymo One	Austin, TX	Fully driverless commercial	Limited hours	Live (expanding)
Zoox	Las Vegas, NV (strip area)	Pilot / invite-only	Limited	Pilot
Zoox	Foster City, CA	Testing w/ safety driver	Daytime	Testing
Cruise	Phoenix, Dallas, Houston	Testing w/ safety driver	Limited	Rebuilding (no driverless service)

If you live outside these metros, you cannot currently hail a driverless robotaxi. Announced expansions to additional cities remain in testing phases and should not be confused with live commercial service.

How Much Does a Robotaxi Cost vs. Uber and Lyft?

For markets where Waymo One operates commercially, pricing is comparable to — and sometimes slightly below — standard UberX or Lyft fares, but not dramatically cheaper. Waymo has used promotional pricing to build ridership, but base rates are in the same ballpark:

Typical Fare Comparison — 5-Mile Urban Trip (2026)

Service	Estimated Fare	Surge Pricing?	Tipping?
Waymo One	$10–$16	Limited surge model	In-app option (not expected)
UberX	$12–$20	Yes — can 2-3x	Expected (~15-20%)
Lyft	$11–$19	Yes — Prime Time	Expected (~15-20%)

To use Waymo One, download the Waymo app (available on iOS and Android), create an account, set your pickup and dropoff within the service area, and confirm. Wait times in Phoenix and San Francisco typically range from 3 to 10 minutes depending on demand and time of day. The experience is remarkably ordinary — which is arguably the point. The vehicle arrives, unlocks, plays a brief safety video on first ride, and navigates to your destination without a human at the wheel.

The Technology Stack: How Robotaxis See the Road

Level 4 robotaxis use a fundamentally different technology approach than consumer Level 2 driver-assistance systems. Understanding this helps explain both their capabilities and their limitations.

Regulatory Landscape: Who Approves Robotaxi Operations?

Robotaxi deployment in the U.S. requires approval from multiple regulatory bodies, and this patchwork is a key reason expansion has been slow:

• State level: Each state sets its own rules for testing and commercial deployment of autonomous vehicles. California's Department of Motor Vehicles (DMV) and Public Utilities Commission (CPUC) both play roles — the DMV issues testing permits, while the CPUC authorizes commercial passenger service. Arizona has taken a permissive approach with minimal regulatory barriers, which is why Phoenix became Waymo's first commercial market. Texas passed legislation in 2023 enabling autonomous vehicle operations with specific insurance and reporting requirements.
• Federal level: NHTSA sets vehicle safety standards (FMVSS) and issues the crash-reporting Standing General Order, but has not yet created a comprehensive federal framework for Level 4 commercial deployment. The NHTSA AV safety portal tracks reported incidents.
• Local level: City governments and transportation authorities often negotiate directly with operators on issues like curb management, emergency responder protocols, and accessibility requirements.

The Cruise suspension in late 2023 demonstrated how quickly regulatory approval can be revoked. California's DMV suspended Cruise's deployment permit within days of the San Francisco incident, and the company voluntarily paused nationwide operations. This regulatory leverage is significant: operators know that a serious safety failure can shut down their entire business overnight.

Current Limitations: Where Robotaxis Cannot Go

Level 4 robotaxis are impressive within their operational domain, but that domain is tightly constrained. Understanding these limitations is critical for anyone considering using the service or evaluating safety claims:

Putting the Crash Data in Perspective

So, how many self-driving cars have crashed? The honest answer requires breaking the question into parts:

1. Level 4 robotaxis (Waymo, Zoox): Dozens of minor incidents have been reported to NHTSA across millions of driverless miles. The vast majority are low-speed contacts — often being rear-ended by human drivers at stoplights, a scenario where the robotaxi is the victim, not the cause. Waymo's peer-reviewed research indicates injury-causing crash rates significantly below human benchmarks in comparable urban environments.
2. Level 2 driver-assist (Autopilot, FSD, Super Cruise): Thousands of incidents in NHTSA's database, but these involve human-supervised systems on millions of consumer vehicles. The driver, not the system, bears legal responsibility. NHTSA has opened specific investigations into patterns of emergency vehicle collisions involving Tesla Autopilot.
3. The Cruise incident: One high-profile event in October 2023 — a pedestrian struck by a human-driven vehicle and then run over and dragged by a Cruise robotaxi — resulted in a nationwide shutdown. This single incident had outsized impact on public perception and regulatory posture, demonstrating that even one serious event can reshape the entire industry.

The comparative safety case for robotaxis rests on a straightforward argument: eliminating the impaired, distracted, or fatigued human driver should reduce crash rates. Early data from Waymo's millions of driverless miles supports this hypothesis, but the sample size remains small relative to the trillions of miles driven by humans annually. Drawing definitive conclusions requires more data, more cities, and more years of operation.

Frequently Asked Questions

Are robotaxis safer than human drivers?

Early data from Waymo suggests yes, within their operational domain. Waymo's 2024 peer-reviewed study reported an 85% reduction in injury-causing crashes compared to human benchmarks in comparable urban driving. However, robotaxis operate in favorable conditions (good weather, mapped roads, moderate speeds) and the total mileage sample is still small. A definitive answer requires years more data across more environments.

Is Tesla FSD a self-driving robotaxi?

No. Tesla Full Self-Driving (Supervised) is an SAE Level 2 driver-assistance system. Despite the "Full Self-Driving" name, the driver must remain fully attentive, keep hands on the wheel (or demonstrate engagement), and is legally responsible for the vehicle at all times. Tesla has discussed launching a robotaxi service, but as of early 2026, no driverless commercial Tesla service exists. Any claims that FSD is "self-driving" conflate marketing language with SAE automation levels.

What happens if a robotaxi gets into a crash with no one inside?

Waymo vehicles carry commercial insurance and are programmed to pull over safely after a collision, contact emergency services if needed, and notify Waymo's remote operations team. The vehicle's sensor data provides a complete record of the incident. In most reported robotaxi crashes, the robotaxi was rear-ended by a human driver — the same most-common crash type in human-only traffic.

Why did Cruise shut down, and is it back?

In October 2023, a Cruise robotaxi in San Francisco ran over and dragged a pedestrian who had been thrown into its path by another vehicle. The incident — and Cruise's initial handling of the event, including incomplete information shared with regulators — led California's DMV to suspend its permit. Cruise voluntarily paused all driverless operations nationwide. As of early 2026, Cruise has resumed testing with human safety drivers in select cities but has not restored fully driverless commercial service.

Can I use a robotaxi in any city?

No. Fully driverless commercial robotaxi service is currently available only in Phoenix, San Francisco, Los Angeles, and Austin through Waymo One, with limited pilot programs from Zoox in Las Vegas. If you are outside these metros, robotaxi service is not available to you regardless of app availability.

How do robotaxis handle bad weather?

They largely don't. Heavy rain, snow, and fog degrade LiDAR and camera performance. Waymo may reduce service areas, increase wait times, or pause operations entirely during severe weather. This is one reason robotaxi deployment has prioritized Sun Belt cities with mild climates. Winter-weather robotaxi service in northern U.S. cities is not expected for several years.