Tesla Autopilot Crashes

Understanding Tesla Autopilot Crash Data

Tesla's Autopilot system, classified as a Level 2 Advanced Driver Assistance System (ADAS), has been involved in 3,550 reported incidents according to NHTSA data. These incidents range from minor fender benders to fatal collisions, and they represent the largest single dataset of autonomous driving incidents from any manufacturer.

Tesla accounts for the majority of ADAS incident reports, which reflects both its market dominance in the electric vehicle space and NHTSA's reporting requirements for vehicles equipped with Level 2 automation. It's important to note that higher incident counts don't necessarily mean less safe systems — Tesla has significantly more vehicles on the road with active ADAS features compared to competitors.

Of the 3,550 total incidents, 57 involved fatalities and 202 involved injuries. The remaining 3,291 incidents resulted in no reported injuries. The average pre-crash speed across incidents where speed data is available was approximately 42 mph.

Severity Distribution

Understanding the severity distribution of Tesla crashes provides important context:

• No injury: 3,291 incidents — the vast majority of Tesla Autopilot crashes are relatively minor, resulting in property damage only.
• Injury: 202 incidents involved at least one injury, ranging from minor to severe.
• Fatal: 57 incidents resulted in at least one death. Fatal crashes disproportionately involve highway speeds and collisions with stationary objects.

Common Crash Patterns

Analysis of the incident database reveals several recurring crash patterns in Tesla Autopilot incidents:

• Rear-end collisions: The most common type, including both striking vehicles ahead and being struck from behind (sometimes after phantom braking events).
• Stationary object strikes: Autopilot has repeatedly failed to detect and brake for stopped vehicles, including parked emergency vehicles with active lights.
• Road departure: Single-vehicle crashes where the vehicle leaves the roadway, often on curves or at lane transitions.
• Intersection incidents: As FSD expands to city streets, intersection-related crashes including red light violations are an emerging pattern.
• Pedestrian and cyclist strikes: Less common but among the most serious, raising questions about the vision system's ability to detect vulnerable road users.

Geographic Distribution

Tesla Autopilot crashes span 51 U.S. states, with concentration in states with high Tesla ownership. Top states for incidents:

• CA: 1212 incidents
• TX: 409 incidents
• FL: 278 incidents
• NJ: 124 incidents
• GA: 111 incidents

California consistently leads in Tesla incidents, reflecting both the state's large Tesla market share and its comprehensive reporting requirements. For state-by-state regulatory context, see our self-driving car laws guide.

ADS vs ADAS Classification

Tesla's incidents split into two categories: 18 classified as ADS (Automated Driving System) and 3532 as ADAS (Advanced Driver Assistance System). The ADS classification typically indicates the vehicle was operating with higher autonomy — likely FSD in a more autonomous mode — while ADAS covers standard Autopilot operation.

This distinction matters for understanding the evolution of Tesla's autonomous driving program. As FSD deployment expands, the ratio of ADS to ADAS incidents is expected to shift, bringing new types of crash scenarios associated with urban driving, intersection navigation, and complex traffic situations.

Data Sources and Methodology

All incident data on AutoPilotWatch comes from federal sources:

• SGO Reports: NHTSA's Standing General Order 2021-01 requires manufacturers to report any crash involving vehicles equipped with Level 2+ autonomous features within strict timelines.
• Consumer Complaints: 2,482 complaints filed directly by consumers through NHTSA's complaint system.
• Investigation Records: Data from NHTSA's Office of Defects Investigation, including Preliminary Evaluations and Engineering Analyses.

It's important to note that SGO reporting captures crashes in equipped vehicles regardless of whether the autonomous features were active. This means some reported crashes may have occurred during manual driving in vehicles that have Autopilot capability.