People killed on the road since the Netherlands granted FSD (Supervised) an EU type approval on 10 April 2026

On 10 April 2026 the Dutch vehicle authority RDW issued a European type approval for Tesla's FSD (Supervised), valid provisionally in the Netherlands only. Because the system does not fit the current UN rulebook for driver-assistance systems (UN R171), Tesla asked for exemptions under Article 39 of Regulation (EU) 2018/858. Extending the approval to the whole EU needs an implementing act from the Commission, which needs a qualified-majority vote of member states in the Technical Committee on Motor Vehicles (TCMV). Five months later that vote has not been held. Four more states recognised the Dutch approval on their own; twenty-two have not.

The counter takes each of those twenty-two states' 2025 road-death total, spreads it evenly across the year, and adds it up from the start date to now. States drop out of the count on the day they permit the system. Seasonality is ignored, so the figure runs slightly low in summer and slightly high in winter. It is an estimate of a real toll, not a tally of identified crashes.

The two derived figures are where the argument lives. Tesla says FSD (Supervised) is 4.1 times less likely to crash than a manually driven Tesla, measured across 100 million km in the five approving states. The first figure assumes that ratio held for every car and that fewer crashes meant proportionally fewer deaths. The second figure asks a narrower question: of the people who died, how many were in a crash involving a Tesla that could have been running FSD (Supervised) had the vote passed? Take-up is deliberately left out: approval is what drives it, so every capable car is assumed to have the system. You can change every factor in the assumptions panel.

Germany, France, Italy and Spain alone hold 58% of the EU's population. No qualified majority exists without at least three of them. France said on 22 July it would not adopt the Dutch approval nationally and wants EU-wide sign-off with changes first. Germany's KBA is running its own evaluation. Sweden's transport agency has asked the committee to vote no unless the speed-limit offset is removed.

The committee meets only every few months. It heard the Dutch presentation on 5 May, discussed the file again on 30 June without voting, and the next slot on which a vote is expected is 6 October 2026. If that passes, the Commission adopts the act and the Dutch approval becomes valid across the Union. If it fails or slips, the next realistic window is December.

The one number that connects Tesla's evidence to European roads is the 4.1× ratio, and it deserves a careful reading. It compares FSD-engaged kilometres with manually driven kilometres in Teslas, using Tesla's own telemetry, with a collision defined as a speed change of at least 8 km/h or an airbag deployment, and the system counted as engaged if it was active within five seconds of impact. It is a ratio of collisions, not of deaths. Tesla publishes no fatality comparison.

The dashboard's regional comparison is also worth knowing. European Teslas driven manually already crash far less than North American ones (about 40–45% fewer major collisions per mile, by road class). Against that safer European baseline, North American FSD data shows no measurable advantage on highways and local roads, a 16% advantage on urban collectors, and a 43% advantage on arterials. The 4.1× figure comes from later, EU-only customer data that Tesla has not yet broken down by road class in public.