Is Tesla Autopilot Safe?

The Short Answer

The safety of Tesla Autopilot is complicated and cannot be reduced to a simple yes or no. The NHTSA data shows that Tesla has reported 3,550 incidents involving Autopilot or FSD-equipped vehicles, resulting in 57 fatalities and 202 injuries. This gives Tesla a fatality rate of 1.6% per reported incident, compared to the overall industry rate of 1.0%.

However, these numbers alone don't answer the safety question. Tesla's massive fleet size, the types of roads where Autopilot operates, and differences in reporting requirements all affect the data. The real question — does Autopilot make driving safer per mile? — remains unanswered by public data.

What the Data Shows

Tesla's incident data must be interpreted carefully. Several important factors affect the numbers:

• Fleet size matters: Tesla has more ADAS-equipped vehicles on U.S. roads than any other manufacturer. More vehicles means more incidents in absolute terms, even if the per-mile rate is comparable or better.
• Reporting requirements: NHTSA requires reporting of all crashes involving vehicles with Level 2+ systems, whether or not the system was active. This inflates the raw count.
• Speed profile: Tesla Autopilot primarily operates at highway speeds (avg 42 mph in incident data), where crashes tend to be more severe than the low-speed urban incidents typical of ADS providers like Waymo.
• System engagement varies: Not all reported incidents occurred while Autopilot was actively engaged. The data includes cases where the system was available but not necessarily controlling the vehicle.

Severity Breakdown

Looking at the severity distribution provides more nuance than raw incident counts:

• No injury: 3,291 incidents (93%) resulted in no reported injuries — the vast majority of Tesla incidents are property-damage-only fender benders.
• Injuries: 202 incidents involved at least one injury, giving an injury rate of 5.7%.
• Fatalities: 57 incidents involved at least one death, representing a 1.6% fatality rate.

The high proportion of no-injury incidents is significant — it suggests that most Autopilot-involved crashes are relatively minor. However, the fatalities that do occur often involve high-speed impacts, including collisions with stationary objects, emergency vehicles, and median barriers.

Compared to Other Systems

Comparing Tesla to other manufacturers is difficult because they operate in fundamentally different domains. Waymo's vehicles operate at low speeds in geo-fenced urban areas. Tesla's Autopilot covers highway driving across the entire U.S. Comparing raw incident counts is misleading without normalizing for miles driven, which neither company makes publicly available for independent verification.

Tesla's own safety reports claim that Autopilot-engaged driving is safer than average human driving when measured by crashes per mile. However, these claims have been disputed by researchers who note methodological issues with Tesla's comparison baseline and the difficulty of accounting for the types of roads where Autopilot is used (primarily divided highways, which are already safer than average).

For a detailed side-by-side breakdown, see our Tesla vs Waymo safety comparison.

Consumer Complaints

Beyond crash reports, 2,482consumer complaints have been filed with NHTSA about Tesla's autonomous features. Common complaint themes include:

• Phantom braking: Sudden unexpected braking on highways — the most common complaint category. See our phantom braking analysis.
• Unintended acceleration: Reports of the vehicle accelerating unexpectedly, though some may be pedal misapplication.
• Lane departure: The vehicle drifting out of its lane or making unexpected lane changes.
• Failure to detect obstacles: The system not recognizing stopped vehicles, pedestrians, or road hazards.

The Injury Rate

Tesla's injury rate per incident is 5.7%. This is influenced by the higher speeds involved in highway incidents. For comparison, ADS providers like Waymo have higher injury rates per incident, but this is partly because their reporting thresholds include minor discomfort complaints from passengers.

The injury rate also reflects the crashworthiness of Tesla vehicles themselves. Teslas consistently receive high safety ratings from NHTSA and IIHS for occupant protection, which likely reduces injury severity in crashes that do occur. The vehicle's structural safety is a separate question from its autonomous driving system safety.

Known Problem Areas

Several specific failure modes have been identified in the NHTSA data and consumer complaints:

• Stationary object detection: Autopilot has repeatedly failed to detect and brake for stationary vehicles, including parked emergency vehicles with flashing lights. This is a fundamental limitation of the vision system at highway speeds.
• Intersection handling: As FSD expands to city streets, traffic signal and stop sign compliance has become a major concern.
• Edge cases: Unusual scenarios — construction zones, temporary lane markings, vehicles pulling trailers — continue to challenge the system.
• Driver complacency: Perhaps the biggest safety issue isn't technical but human. The system's capability encourages over-reliance, and Tesla's branding as "Full Self-Driving" may exacerbate this.

Emergency Vehicle Crashes

One of the most concerning patterns in the Tesla Autopilot data is collisions with stationary emergency vehicles. Multiple incidents have involved Autopilot-engaged Teslas striking parked fire trucks, police cars, and ambulances — often with their emergency lights active. These crashes suggest a fundamental limitation in the vision system's ability to detect and respond to stationary objects at highway speeds.

This pattern was a significant factor in NHTSA's decision to open investigations into Autopilot. The agency documented at least a dozen emergency vehicle collisions that contributed to the PE22-002 investigation. First responders have expressed particular concern, as emergency scenes require vehicles to move over or slow down — something Autopilot has repeatedly failed to do reliably.

Bottom Line

The data does not definitively prove Tesla Autopilot is safe or unsafe. What it shows is that crashes do occur with automation engaged, including fatal ones. Whether the technology reduces overall crash risk compared to unassisted driving remains an open question that requires per-mile exposure data that isn't publicly available. Until then, drivers should treat Autopilot as the Level 2 system it is: a driver assistance tool that requires constant attention.

For a broader perspective on autonomous vehicle safety across all manufacturers, see our guide on how safe self-driving cars really are.