Frequently Asked Questions

Tesla Robotaxi Safety FAQ

Everything you need to know about our methodology, data sources, incident definitions, and how to interpret robotaxi safety statistics

By Kangning Huang · Last updated: February 4, 2026

-- Latest MPI
-- Days to Double
-- Total Incidents
-- Data Through

Methodology & Data Sources

How we calculate miles per incident and where our data comes from

NHTSA SGO 2021-01 is a federal requirement that mandates manufacturers and operators of vehicles equipped with SAE Level 2 ADAS or Levels 3-5 ADS to report certain crashes to NHTSA.

Key reporting requirements:

  • Crashes must be reported if ADS was engaged within 30 seconds of the crash
  • Reports are required for crashes involving any injury, fatality, vehicle tow-away, airbag deployment, or vulnerable road user
  • Reports must be submitted within 1 day for serious incidents, 10 days for others

This is the primary public data source for tracking autonomous vehicle incidents across all manufacturers, including Tesla, Waymo, Cruise, and others.

MPI = Fleet Miles Driven ÷ Number of Incidents

We calculate this for each interval between consecutive incidents:

  1. Fleet size: Daily vehicle count from robotaxitracker.com
  2. Daily miles: 115 miles/vehicle/day (based on Tesla's Q3 2025 disclosure)
  3. Interval miles: Sum of daily fleet miles between consecutive incident dates

This gives us the actual distance driven between each incident, accounting for fleet growth over time.

Tesla disclosed in their Q3 2025 earnings report that their Austin robotaxi fleet was averaging approximately 115 miles per vehicle per day. This is consistent with typical robotaxi utilization rates (compared to ~30 miles/day for private vehicles).

Limitations:

  • This is an average; individual vehicles may drive more or less
  • Utilization may vary by day of week or time of year
  • If Tesla updates this figure, we will adjust our calculations

Even with some uncertainty in this estimate, the exponential improvement trend (R² = --) is robust to reasonable variations in daily mileage assumptions.

Austin is the only location where Tesla operates unsupervised Level 4 autonomous driving. This is the only deployment that:

  • Requires mandatory NHTSA SGO incident reporting
  • Operates without a safety driver ready to intervene
  • Represents true autonomous vehicle performance

The Bay Area fleet operates with safety drivers (Level 2) as required by California law. Different reporting requirements apply, and including that data would conflate different operational modes.

We track daily fleet size using multiple sources:

  • Primary: robotaxitracker.com, which monitors Tesla's API for active vehicles
  • Secondary: Tesla earnings calls and press releases
  • Cross-reference: News reports and third-party tracking

The current Austin fleet is approximately -- unsupervised vehicles.

Reporting Lag & Data Freshness

Understanding when incidents are reported and how this affects the data

Reporting lag is the delay between when an incident occurs and when it appears in NHTSA's public database. Under SGO 2021-01:

  • Serious incidents: 1 day to report
  • Other incidents: 10 days to report
  • NHTSA processing: Additional days before public posting

Why it matters: If you count "miles driven through Sunday" but incidents haven't been reported yet, you may be calculating MPI with an incomplete incident count, making safety appear better than it actually is.

This tracker displays both "Data Through" (latest mileage date) and "Latest Report" (most recent incident report date) so you can assess data freshness.

Current data status:

  • Fleet miles through: --
  • Latest incident report: --
  • Last NHTSA check: Updated daily

We check NHTSA's SGO database daily and update incident data as new reports are published. The "current streak" metric (miles without an incident) is inherently provisional until sufficient time has passed for any incidents to be reported.

The current streak (-- miles) should be treated as provisional, especially for the most recent 7-14 days.

Guidance:

  • If the last incident was 14+ days ago: streak is likely accurate
  • If the last incident was 7-14 days ago: treat with caution
  • If the last incident was <7 days ago: expect possible updates

The completed interval MPI values (between known incidents) are the most reliable metrics, as both the start and end incidents have been reported.

Different sources may show different incident counts due to:

  • Timing: They may have checked NHTSA at a different time
  • Scope: They may include Bay Area (supervised) incidents
  • Definition: They may count incidents differently (e.g., including disengagements)

This tracker specifically counts only Austin (unsupervised) incidents from NHTSA SGO ADS reports. Our methodology is documented and consistent.

Comparisons to Human Drivers

Why "robotaxi vs human" comparisons are more complicated than they appear

Comparing robotaxi incident rates to human crash rates involves several mismatches:

  1. Reporting threshold: NHTSA SGO captures incidents that human drivers would never report to police (minor parking lot bumps, low-speed contacts)
  2. Operating environment: Robotaxis operate in geofenced urban areas; human stats include highways, rural roads, and all conditions
  3. Fault attribution: SGO reports include incidents where the robotaxi was not at fault (e.g., being rear-ended)
  4. Severity: Most robotaxi incidents are low-severity; human crash stats weight differently by severity

These factors can make comparisons unfair in either direction, depending on what you're measuring.

Several benchmarks exist, each measuring something different:

Baseline MPI What it measures
Police-reported crashes ~500,000 Crashes serious enough for police report
Insurance claims ~300,000 Crashes where insurance was involved
All crashes (estimated) ~100,000-200,000 Including unreported minor incidents
Urban taxi/rideshare Unknown Most relevant but not publicly tracked

The insurance claim baseline (~300,000 MPI) is arguably the most appropriate comparison for robotaxis since SGO captures incidents that would result in claims.

An ideal comparison would control for:

  • Geography: Same city, same streets
  • Time: Same hours of operation (e.g., no late-night drunk driving hours)
  • Vehicle type: Similar vehicle class
  • Reporting standard: Same incident threshold
  • Fault: Only at-fault incidents for both

Unfortunately, no public dataset exists for "Austin urban taxi/rideshare incident rate using SGO-equivalent reporting." The closest would be comparing to Uber/Lyft incident rates in Austin, but that data isn't public.

What we can say: Tesla's trend shows rapid improvement. Whether the current rate is "better" or "worse" than humans depends heavily on what comparison you choose.

Waymo reports approximately 1,000,000+ miles per incident based on their published safety data and the Swiss Re study.

Tesla's current MPI (-- miles) is approximately --x less than Waymo's reported figures.

Important caveats:

  • Waymo has been operating for longer and in more cities
  • Waymo uses different sensor suites (LiDAR + cameras vs. Tesla's vision-only)
  • Waymo's data is self-reported; Tesla's comes from NHTSA SGO
  • Tesla's rapid improvement rate (doubling every ~-- days) could narrow this gap

This is a fair criticism. Tesla's Austin robotaxis operate in a limited Operational Design Domain (ODD):

  • Urban/suburban streets only
  • Specific geofenced areas
  • Good weather conditions
  • Mapped routes

National human crash statistics include:

  • Highway driving (higher speeds, different risks)
  • Rural roads (less infrastructure, wildlife)
  • All weather conditions
  • All driver conditions (fatigue, impairment)

The geofence cuts both ways: Urban areas have more pedestrians, cyclists, and complex intersections than highways. The "easy" vs "hard" comparison isn't straightforward.

Incident Definitions & Fault

What counts as an incident and who's at fault

An "incident" on this site means a crash event that appears in NHTSA SGO 2021-01 ADS reports for Tesla's Austin robotaxi program.

This includes:

  • Any crash where ADS was engaged within 30 seconds
  • Crashes involving injury, tow-away, airbag deployment, or vulnerable road users
  • Minor contacts that meet SGO reporting thresholds

This does NOT include:

  • Near-misses or close calls
  • Disengagements without contact
  • Bay Area (supervised) incidents

Yes. SGO 2021-01 has a lower reporting threshold than police reports. It can capture:

  • Low-speed parking lot contacts
  • Minor fender benders with no injuries
  • Incidents that human drivers would handle without police

This is a legitimate criticism of direct MPI comparisons. A robotaxi might have a "worse" MPI than police-reported human stats while actually being safer, simply because more incidents are being reported.

Counterpoint: SGO doesn't capture all minor incidents — only those meeting specific criteria. Many true minor contacts may still go unreported.

SGO reports do not include fault determination. This is a significant limitation of the public data.

What we know:

  • Some incidents are clearly not the robotaxi's fault (e.g., being rear-ended at a stop)
  • Tesla redacts crash narratives, making fault analysis difficult
  • Without video or detailed reports, we cannot reliably categorize fault

Why this matters: If a significant portion of incidents are other vehicles hitting the robotaxi, the "true" at-fault MPI would be much better than the overall MPI.

We are exploring ways to provide provisional fault categorization based on available evidence, but this is inherently uncertain.

Based on available SGO data for Tesla's Austin fleet:

  • Fatalities: 0
  • Serious injuries: 0
  • Minor injuries: Rare
  • Property damage only: Most incidents

The lack of serious injuries or fatalities is notable, though the sample size is still relatively small. This suggests that even when incidents occur, they tend to be low-severity.

Yes. Tesla redacts the narrative section of SGO crash reports, preventing the public from knowing specific details about how crashes occurred.

What is NOT redacted:

  • Incident dates
  • ADS engagement status
  • Injury/fatality status
  • Vehicle identification

This is a legitimate transparency concern. Waymo, by comparison, provides more detailed incident descriptions. However, the core metrics tracked here (MPI, trend) are calculable from non-redacted data.

Supervision & Interventions

Understanding the role of safety monitors and remote operators

The situation is mixed:

  • Some vehicles: Operate fully unsupervised (no human in vehicle)
  • Other vehicles: Still have safety monitors present who can intervene

Tesla has been progressively removing safety monitors from Austin vehicles as confidence increases. The exact split is not publicly disclosed.

This is a key confound. If safety monitors intervene to prevent crashes, the observed MPI reflects "FSD + human backup" rather than "FSD alone."

What we don't know:

  • Intervention rate (how often monitors take over)
  • Which specific incidents involved monitored vs unmonitored vehicles
  • Whether interventions are preventing crashes or just providing comfort

As Tesla removes more monitors, the data will increasingly reflect true autonomous performance. If MPI stays strong with fewer monitors, that's a positive signal.

Tesla has remote operators who can assist vehicles that get stuck or need guidance. However:

  • Remote operators cannot directly control the vehicle (unlike Waymo's teleops)
  • They can provide route guidance or approve certain maneuvers
  • This is not real-time driving assistance

The extent to which remote assistance affects safety performance is unknown.

Fleet-wide removal of safety monitors is expected when:

  • Safety metrics: Sustained MPI at or above human-driver benchmarks (300,000-500,000 miles)
  • Geographic consistency: Performance maintained across expansion cities
  • Regulatory approval: Permits obtained in states that require them (e.g., California)
  • Insurance/liability: Appropriate coverage in place

This tracker focuses on the quantitative signals that predict when removal becomes operationally rational.

Chart Interpretation & Statistics

How to read the charts and understand the statistical methods

We use a logarithmic (log) scale because:

  • Exponential trends appear as straight lines on a log scale, making it easier to see whether improvement is consistent
  • Percentage changes are visually equal: A doubling from 10K to 20K looks the same as 50K to 100K
  • Large variations are visible: Without log scale, early low values would be squished at the bottom

What to watch for: On a log scale, linear improvement looks like it's slowing down. Our trend line shows the data is consistent with exponential (not linear) improvement.

The trend line is an exponential regression of the form:

MPI = a × e^(b × days)

This model assumes safety improves at a constant percentage rate over time (like compound interest). Key outputs:

  • Doubling time: -- days for MPI to double
  • R²: -- (how well the model fits the data)

An R² close to 1.0 means the exponential model explains most of the variation in the data.

Yes, somewhat. With a small number of data points (currently -- incidents), each new incident affects the trend.

How we address this:

  • We report the R² value so you can see fit quality
  • We show all data points, not just the trend line
  • The doubling time is a derived metric — if the trend breaks, doubling time will change significantly

As more incidents occur, the trend becomes more robust. Early-stage data naturally has higher uncertainty.

Some people argue log scales "compress" the gap between Tesla and human baselines. This is both true and misleading:

  • True: On a log scale, 50K vs 500K looks "closer" than on a linear scale
  • Misleading: The log scale is the correct way to visualize exponential data — using a linear scale would misrepresent the improvement rate

We show the numeric values clearly. Tesla's latest MPI is -- miles. Human baseline (police-reported) is ~500,000 miles. You can judge the gap yourself.

The exponential improvement trend could break if:

  • Fleet expansion: New cities/vehicles might have different performance
  • Software changes: A new FSD version could perform differently
  • Operational changes: Removing more safety monitors might reveal hidden issues
  • Statistical regression: Early data might have been unusually lucky/unlucky

We update daily and will note any significant deviations from the trend. If MPI stops improving exponentially, the doubling time estimate will increase or become undefined.

Common Misconceptions & News Coverage

Addressing frequent misunderstandings and media claims

These reports are mathematically correct but contextually incomplete:

  • They divide total incidents by total miles since launch (simple average)
  • This yields ~-- MPI, which is indeed worse than human rates

What they miss:

  • The data shows exponential improvement over time
  • The latest interval (-- miles) is much better than the average
  • Judging by simple average is like grading a student by lifetime GPA instead of their most recent semester

The question isn't "what was Tesla's average safety?" but "what is Tesla's current safety and trajectory?"

Different numbers arise from:

  • Different dates: Checking NHTSA at different times
  • Different scope: Some include Bay Area (supervised) incidents
  • Different definitions: Some count all ADS incidents (including non-crash events)
  • Outdated data: Articles written weeks/months ago

This tracker uses consistent methodology: Austin unsupervised ADS crashes from NHTSA SGO, updated daily.

This site aims to be data-driven and transparent:

  • All data comes from public sources (NHTSA SGO, robotaxitracker.com)
  • Methodology is fully documented
  • Limitations and caveats are explicitly stated
  • The code is open source

What the data shows: Tesla's robotaxi safety is improving exponentially but currently lags behind Waymo and human baselines. We present both the positive trend and the current gap.

If you find errors or methodological issues, please report them on GitHub.

This is a personal decision that depends on your risk tolerance and how you weigh different factors:

  • Current data: Tesla's robotaxis have had -- incidents with no fatalities or serious injuries
  • Trend: Safety is improving rapidly
  • Comparison: Still below human and Waymo benchmarks on raw MPI
  • Severity: Robotaxi incidents tend to be low-severity

We provide the data; you decide what it means for you.

See the Latest Data

Track Tesla robotaxi safety metrics in real-time with our interactive dashboard

View Safety Tracker