13:00 30-07-2026

Tesla FSD Supervised in Europe: zero collisions in 41.9 million motorway kilometres

Tesla has published its first five-country European safety report for FSD Supervised. Zero motorway collisions in 41.9 million km, but off the motorway the advantage drops to 2.8 times.

Tesla has for the first time pulled together FSD Supervised safety data from five European countries at once, and the figures it picked are loud ones. Between 10 April and 26 July 2026, cars with the system switched on were involved in 5.2 times fewer collisions than Teslas driven manually. In total, the European fleet covered about 65.7 million km with FSD engaged.

The sample covers the Netherlands, Lithuania, Estonia, Denmark and Belgium — the five EU countries where FSD Supervised is already permitted. Tesla started counting in each of them on the day of local approval, so the observation windows differ in length. Despite the Full Self-Driving name, the car does not become autonomous: the driver has to keep watching the road and be ready to step in.

The showiest part of the report is the motorway data. With FSD active, 41.9 million km passed without a single recorded collision. Over the same period the control group racked up 88 crashes across 429.3 million km — roughly one every 4.9 million km. That is where the claimed 8.5-fold advantage comes from: the crash-free distance on FSD turned out to be that many times longer than the average interval between collisions under manual control.

Off the motorway the gap is far more modest: 6 collisions over 23.8 million km against 368 over 514.5 million km for manual driving — an advantage of just 2.8 times. It is the mix of the two modes that produces the headline 5.2.

Tesla also said that with FSD engaged the automatic emergency braking system intervened 13.4 times less often. Harsh acceleration was recorded 9.2 times less often, hard braking 7.1 times less often, and emergency swerves 7.6 times less often.

One detail usually gets lost in the retellings: the control group is not an abstract average driver but the same Teslas with the same set of active safety features, differing only in whether FSD was on. The company’s American reports measure its technology against the entire driving population, and that benchmark is noticeably softer. In this sense the European sample is tougher on Tesla itself.

Even so, the raw ratios do not prove that the system is objectively that many times safer than a human. Route structure, traffic density and weather stay off the record, and owners who pay for FSD and switch it on first are by definition not an average slice of the fleet.

Tesla discloses its counting criteria only in part. Under the Vehicle Safety Report methodology, collisions are split into serious ones — with airbags or other non-reversible restraints deploying — and minor ones, defined by a change-in-velocity threshold. A crash is attributed to the system if FSD was active at any point in the five seconds before impact; at the American regulator NHTSA the equivalent window is six times wider, at thirty seconds. The data is gathered automatically through telemetry, and collision figures are updated quarterly on a rolling 12-month basis.

The absence of crashes over 41.9 million km remains a meaningful result, but statistically it does not mean zero risk. All the more so because the original Dutch permit was issued by the RDW after a year and a half of its own testing rather than off the maker’s slide decks, and it is provisional: the system went through under UN Regulation No. 171 for driver assistance systems plus an exemption in European type-approval law.

The other four countries simply recognised the Dutch decision, and only Belgium ran additional testing of its own. Germany, France, Italy and Spain have granted no approval and are waiting for an EU-wide one: the vote in the European technical committee is not expected before October.

The backdrop is an awkward one. In June, Reuters reported that Tesla had presented regulators in the Netherlands and Sweden with statistics that independent safety researchers called marketing, while the European Transport Safety Council urged transport ministers not to recognise the Dutch permit until questions about the system are answered publicly. The new European data strengthens Tesla’s position, but a final verdict will require independent checks on comparable roads and a single methodology for counting crashes.