Self-driving car safety depends on more than the vehicle alone—it’s shaped by human behavior, roadway design, and real-world conditions shared by everyone on the road. This page uses U.S. and global crash data, including how alcohol-impaired and distracted driving affect outcomes, and how factors like adverse weather can challenge perception. It also links those findings to the technical and regulatory building blocks behind automated driving.
Key Takeaways
- In 2022, 65% of fatal crashes involved passenger vehicles in the U.S. (share of fatalities by vehicle type)
- 2.51% of all U.S. drivers involved in fatal crashes in 2022 had a BAC of 0.08 g/dL or higher
- Alcohol-impaired driving fatalities were 31% of total U.S. traffic fatalities in 2022
- The NHTSA Fatality Analysis Reporting System (FARS) covers crashes occurring on public roads in the United States
- UNECE Regulation No. 152 requires vehicles to be equipped with an event data recorder (EDR) for certain data categories
- UNECE Regulation No. 157 (LDW/ELK-related requirements) entered into force for type approvals with specified scope and technical requirements
- Cruise reported a 2023 rate of 0.80 disengagements per 1,000 miles for its self-driving program (reported metric)
- Waymo reported 2023 disengagements per 1,000 miles of 1.06 for its robotaxi operations (reported metric)
- Tesla’s Autopilot reportedly accounted for 2023 miles per incident using its disclosed metric in its safety report (incident-based disclosure)
- 45% of road deaths in the European Union (EU) in 2022 occurred on roads outside urban areas (rural roads)
- 90% of all road traffic deaths are road users with no crash protection (pedestrians, cyclists, and motorcyclists) in the European Union (EU) (share of fatalities)
- 1,126,000 police-reported crashes involving distracted driving occurred in the United States in 2022 (estimated from NHTSA’s annual distracted driving crash data)
- The U.S. Federal Motor Vehicle Safety Standards (FMVSS) for advanced driver assistance systems include performance requirements for lane keeping support and related technologies (count of named FMVSS/requirements sets applicable to ADAS within FMVSS 111/others in the FMVSS framework)
- In the EU, the General Safety Regulation (Regulation (EU) 2019/2144) requires new vehicles to include Intelligent Speed Assistance (ISA) starting from 2022/2024 timelines (starting phase-in year: 2022)
- ISO 26262 defines Automotive Safety Integrity Levels (ASILs) A through D (4 discrete integrity levels)
Even with progress in ADAS and testing, most road deaths still come from human factors, especially alcohol, distraction, and unprotected road users.
Related reading
01 · Category
Safety Performance9 stats
01
Cruise reported a 2023 rate of 0.80 disengagements per 1,000 miles for its self-driving program (reported metric)
02
Waymo reported 2023 disengagements per 1,000 miles of 1.06 for its robotaxi operations (reported metric)
03
Tesla’s Autopilot reportedly accounted for 2023 miles per incident using its disclosed metric in its safety report (incident-based disclosure)
04
A 2021 peer-reviewed study estimated that advanced driver assistance systems could reduce rear-end crashes by up to 27% under certain conditions
05
A 2020 meta-analysis found automatic emergency braking reduced collision rates with lead vehicles by about 38%
06
A 2019 study on lane-keeping assistance systems found a reduction in lane departure crashes by 15% (estimate from observed outcomes)
07
A 2018 peer-reviewed study on distracted driving using in-vehicle automation reported a decrease in driver reaction time variability by 12% when ADAS engaged (experimental measure)
08
A 2022 paper in Transportation Research Part C quantified that object detection models can exhibit higher false positives in adverse weather, with reported increases of up to 2.3x for certain classes
09
The RAND report 'Validation for Automated Driving Systems' (2020) proposed a safety case framework based on scenarios and evidence accumulation, with measurable completeness criteria
Interpretation
Safety Performance Interpretation
Across these safety performance metrics, the reported disengagement and incident rates cluster around about 0.80 to 1.06 per 1,000 miles while peer reviewed research shows that well targeted automated driving features can cut specific crash types by roughly 15% to 38%, suggesting measurable safety gains are already showing up as systems move from constrained performance to real world operation.
02 · Category
Regulation & Standards5 stats
01
The NHTSA Fatality Analysis Reporting System (FARS) covers crashes occurring on public roads in the United States
02
UNECE Regulation No. 152 requires vehicles to be equipped with an event data recorder (EDR) for certain data categories
03
UNECE Regulation No. 157 (LDW/ELK-related requirements) entered into force for type approvals with specified scope and technical requirements
04
UNECE Regulation No. 160 sets requirements for advanced driver assistance systems (ADAS) to be designed to ensure safe operation and includes performance criteria
05
As of 2024, the ISO 26262 standard defines a functional safety framework for road vehicles with an approach to automotive safety lifecycle
Interpretation
Regulation & Standards Interpretation
Across Regulation and Standards, global rules and safety frameworks are tightening alongside national reporting, with NHTSA’s FARS covering public road fatalities while UNECE standards now mandate vehicle event data recording through Regulation No. 152 and require safe ADAS design under Regulation No. 160, supported by ISO 26262’s functional safety framework as of 2024.
03 · Category
Validation & Evidence5 stats
01
The U.S. Federal Motor Vehicle Safety Standards (FMVSS) for advanced driver assistance systems include performance requirements for lane keeping support and related technologies (count of named FMVSS/requirements sets applicable to ADAS within FMVSS 111/others in the FMVSS framework)
02
In the EU, the General Safety Regulation (Regulation (EU) 2019/2144) requires new vehicles to include Intelligent Speed Assistance (ISA) starting from 2022/2024 timelines (starting phase-in year: 2022)
03
ISO 26262 defines Automotive Safety Integrity Levels (ASILs) A through D (4 discrete integrity levels)
04
A 2023 peer-reviewed study reports that adverse weather can increase object detection false positives; in worst-case conditions, certain classes increased by up to 2.3x (reported in the paper)
05
In the U.S., the Cybersecurity and Infrastructure Security Agency (CISA) and NHTSA coordinated on a framework for vehicle safety and security risk management that references safety assurance evidence (version released in 2023; count of risk management principles: 6)
Interpretation
Validation & Evidence Interpretation
For Validation and Evidence, safety frameworks increasingly rely on measurable standards and risk-based assurance, from ISO 26262’s four ASIL levels and the EU’s mandatory intelligent speed assistance to U.S. performance requirements and guidance that incorporates factors like adverse weather detection false positives and coordinated cybersecurity safety validation.
More related reading
04 · Category
Crash & Risk4 stats
01
In 2022, 65% of fatal crashes involved passenger vehicles in the U.S. (share of fatalities by vehicle type)
02
2.51% of all U.S. drivers involved in fatal crashes in 2022 had a BAC of 0.08 g/dL or higher
03
Alcohol-impaired driving fatalities were 31% of total U.S. traffic fatalities in 2022
04
11,258 people were killed in U.S. crashes involving distracted drivers in 2022
Interpretation
Crash & Risk Interpretation
For the Crash and Risk angle, the data show that high risk factors behind deadly crashes are largely tied to real-world driver and road exposure, with 65% of U.S. fatal crashes involving passenger vehicles in 2022 alongside 31% of all traffic deaths stemming from alcohol impairment and 11,258 people killed in crashes involving distracted drivers.
05 · Category
Road Safety Burden3 stats
01
45% of road deaths in the European Union (EU) in 2022 occurred on roads outside urban areas (rural roads)
02
90% of all road traffic deaths are road users with no crash protection (pedestrians, cyclists, and motorcyclists) in the European Union (EU) (share of fatalities)
03
1,126,000 police-reported crashes involving distracted driving occurred in the United States in 2022 (estimated from NHTSA’s annual distracted driving crash data)
Interpretation
Road Safety Burden Interpretation
In the Road Safety Burden context, the fact that 45% of EU road deaths in 2022 happened on rural roads and that 90% of EU road traffic deaths involved road users with no crash protection shows the highest risk is concentrated outside cities and among vulnerable users, while in the US 1,126,000 police-reported distracted driving crashes in 2022 underscores a major avoidable driver-related contributor to crashes.
06 · Category
Industry Overview3 stats
01
The World Health Organization (WHO) estimates 1.19 million road deaths annually worldwide (global annual road traffic death estimate)
02
The OECD International Transport Forum (ITF) reports automated driving is being tested worldwide across multiple regions with regulatory frameworks evolving (count of jurisdictions actively publishing testing frameworks: 20+ as of 2024)
03
In 2022, SAE International published ISO/SAE 21434 cybersecurity engineering for road vehicles and sets requirements for lifecycle cybersecurity engineering (standard publication year: 2022)
Interpretation
Industry Overview Interpretation
With WHO estimating 1.19 million annual road deaths worldwide, the industry is pushing automated driving pilots across multiple regions and strengthening cybersecurity standards like ISO/SAE 21434 in 2022 to address safety risks from both operation and connected vehicle vulnerabilities.
Reference
Cite This Report
This report is designed to be cited. We maintain stable URLs and versioned verification dates. Copy the format appropriate for your publication below.
APA
Karl Becker. (2026, February 13). Self-Driving Cars Safety Statistics. Gitnux. https://gitnux.org/self-driving-cars-safety-statistics
MLA
Karl Becker. "Self-Driving Cars Safety Statistics." Gitnux, 13 Feb 2026, https://gitnux.org/self-driving-cars-safety-statistics.
Chicago
Karl Becker. 2026. "Self-Driving Cars Safety Statistics." Gitnux. https://gitnux.org/self-driving-cars-safety-statistics.
Sources & references
29 datasets cited across this report · attribution is report-level
+11 additional datasets cited (not shown individually)