Tesla FSD (Supervised) in Europe: eight countries have approved it. What would have to change for Romania
Where FSD (Supervised) is approved in the EU, how the Dutch approval and the December vote work, what the safety data says (and its critics), how it copes with Dutch cyclists, and the legal hurdles in Romania: RAR and Article 99 of the Road Code.
In April, Tesla FSD (Supervised) was approved in a single European country. Today it is approved in eight EU member states: the Netherlands, Lithuania, Estonia, Denmark, Belgium, Slovenia, Czechia and, since 29 September, Croatia. Romania is not on the list, and the vote that could make the system legal across the Union has been pushed to December.
We wanted a reference piece rather than another short news item. Below you'll find where FSD is approved and on what legal basis, what the safety data and its critics say, how it copes with Dutch cyclists, and what would actually have to change in Romania. Throughout, we keep verified facts separate from our own interpretation, and we cite sources at every step.
First, the basics: FSD (Supervised) is not autonomous driving. The Dutch authority RDW describes it as a driver-controlled assistance system: the car takes over many tasks, but the driver remains responsible and must be able to take control immediately.
Where FSD (Supervised) is approved in Europe
| Country | Date | Who decided | Source |
|---|---|---|---|
| Netherlands | 10 April 2026 | RDW (provisional approval after more than 18 months of testing) | RDW |
| Lithuania | 20 May 2026 | Recognition of the Dutch approval (Transport Safety Administration) | Reuters, Electrek |
| Estonia | 29 May 2026 | Transpordiamet | Transpordiamet |
| Denmark | 9 June 2026 | Færdselsstyrelsen, after its own review of the documentation | DR |
| Belgium | 10 June 2026 | Flemish mobility minister Annick De Ridder (decision applies nationwide) | EVwire |
| Slovenia | 8 September 2026 | Traffic Safety Agency (AVP), acting as type-approval authority | gov.si |
| Czechia | 21 September 2026 | Ministry of Transport (provisional recognition) | Prague Daily |
| Croatia | 29 September 2026 | State Office for Metrology, acting as type-approval authority | HINA |
A few notes. Electrive and Teslarati calculated that the first seven countries (before Croatia) had about 53 million inhabitants, roughly 12% of the EU population. Germany, France, Italy and Spain have not recognised the approval. Ireland and Italy are in talks with Tesla, and Greece said in May it was preparing a law to the same effect, but none of them has approved the system yet. We covered Germany's position and the EU vote moving to December separately.
According to Electrek, the European subscription costs €99 a month, or €49 for owners who previously bought Enhanced Autopilot.
How approval works: RDW, Article 39 and the December vote
Everything rests on Article 39 of Regulation (EU) 2018/858, the EU's vehicle type-approval framework. Article 39 allows approval of "new technologies or new concepts" that don't fit the existing rules. The condition is that the manufacturer proves, with tests, that safety is at least equivalent. The steps, as the regulation sets them out:
- Provisional national approval (para. 4). Until the Commission decides, a national authority may grant a provisional EU type-approval valid only on its own territory. RDW did this on 10 April. According to Electrek, the approval was granted under UN Regulation R171 (DCAS, driver control assistance systems).
- Acceptance by other states (para. 5). Approval authorities in other member states may accept the provisional approval on their territory, provided they notify the issuing authority in writing. That is what the other seven countries did. The Flemish minister's letter, for example, explicitly cites "Article 39, point 5" and rests on the premise that the driver remains responsible at all times.
- The Commission's decision (para. 3). The Commission authorises or refuses the exemption through an implementing act under the examination procedure. The vote takes place in the Technical Committee on Motor Vehicles (TCMV). According to Reuters and Profit.ro, it takes 55% of member states (at least 15) representing at least 65% of the EU population. If approved, the type-approval becomes valid across the EU.
- If it is refused (para. 7), the provisional approval is revoked six months later. The Danish authority has warned that national recognitions would then fall too.
On 6 October the TCMV discusses the file without a vote (25 minutes on the agenda, according to Profit.ro). The next chance for a vote is the December meeting. Meanwhile, Sweden has asked for the settings that allow driving above the speed limit to be fixed, and Profit.ro reports that a Belgian NGO flagged cases where the system displayed the wrong speed limit.
What it offers at European level, and what we can't claim
What the authority that tested it says
RDW makes no miracle claims, but its conclusion is clear. In its April statement it says that using the system correctly makes a positive contribution to road safety, because the driver is supported in the driving task. RDW also explains that hands don't have to stay on the wheel, but the driver must be able to take over immediately. Sensors check that the driver's eyes are on the road. If the driver is repeatedly inattentive, the system issues warnings and eventually locks itself out temporarily.
After the Dutch parliamentary debate on 16 June, RDW gave more detail on the process:
- it analysed data from 1.8 million kilometres driven in Europe with FSD Supervised and independently checked how Tesla collected it;
- it tested the system for more than 3,000 hours, on test tracks and public roads, including busy urban traffic and extreme weather, in more than 1,000 test runs;
- it has, for now, raised the mandatory reporting frequency from annual to monthly;
- at that point, nearly 40,000 Teslas with FSD Supervised had driven about 24 million km in the Netherlands "without any relevant incidents", in RDW's words.
Tesla's data: big numbers, disputed methodology
Tesla publishes its own safety report. The North American version, cited by Learn My EV and other outlets in July, showed one major collision (airbag deployment) every 5,690,675 miles with FSD engaged. The same fleet driven manually with active safety features managed 2,076,014 miles, and Tesla's estimate of the US average was 698,781 miles. In Europe, according to Profit.ro, Tesla says it has more than 70,000 cars in the countries where FSD is approved and a collision risk four times lower than a human driver's.
These are manufacturer-reported figures, not official statistics. Reuters published an investigation in May and an exclusive in June in which independent traffic-safety researchers describe Tesla's comparisons as misleading. The main objections: new cars are compared with a much older average US fleet, and airbag crashes are compared with a baseline that includes less severe crashes. Learn My EV's analysis shows that comparing FSD with the same Tesla fleet driven manually shrinks the gap from roughly 7x to roughly 2.7x. The European Transport Safety Council (ETSC) told Reuters it was "certainly concerned" about the data presented to Swedish regulators and that such figures should be independently verified. RDW replied that it does not rely on marketing claims but on its own tests and checks.
Our conclusion is cautious: RDW's verdict after 18 months of testing is a stronger argument than any marketing chart. How much FSD reduces crashes in Europe has not yet been independently demonstrated.
Fatigue and long journeys
The European Commission's European Road Safety Observatory (ERSO) notes that police reports attribute only 1-4% of crashes to sleepiness. ERSO also says those reports probably greatly underestimate the problem. An in-depth study it cites found that about 20% of motorway crashes were sleep-related. A system that handles steering, braking and lane changes, while forcing the driver to stay attentive, may reduce the workload on long drives. That is our interpretation, not a measured result for FSD.
Accessibility: where we need to be honest
FSD (Supervised) does not make driving accessible to people who can't drive. It needs a licensed driver who is fit to drive and attentive at all times. It may make driving less tiring for older drivers or for people who spend many hours a day at the wheel, but the responsibility stays entirely with them.
Why it matters for Romania
The European Commission's figures are harsh. According to preliminary 2025 data, Romania had 68 road deaths per million inhabitants, against an EU average of 43. The Commission says Romania remains among the countries with the highest fatality rates in the EU, together with Bulgaria (71) and Croatia (67). For comparison, Sweden has 20, Denmark 23 and the Netherlands 35 (partial data). There is progress, too: Romania's road deaths fell 12% compared with 2024 and 30% compared with 2019, and the Commission lists Romania among the states on track for the 50% reduction target by 2030.
The same release shows that, across the EU, 53% of deaths happen on rural roads and only 8% on motorways (2024 data). FSD (Supervised) is not limited to motorways. RDW tested it in complex urban traffic and on a wide range of road types. Unlike motorway-only hands-off systems, it could be used on the national roads where many serious crashes happen. This is analysis, not a promise: no study has yet measured FSD's effect on Romanian roads. The system has its own limits, from potholes (which a creator who tested FSD in Croatia mentions) to misread speed signs.
There's also a practical argument. Many Romanians drive long distances, between cities or to Western Europe. For planning those trips, we've explained separately how much real-world range matters and how WLTP, CLTC and EPA differ. A system that reduces fatigue on such routes is exactly the kind of technology worth testing seriously here too.
What would have to change in Romania
Here we carefully separate facts from our analysis. We are not lawyers, and a full legal analysis should come from specialists.
The facts
- RAR is Romania's type-approval authority. Government Ordinance (OG) 78/2000 provides that road vehicles may be sold, registered or recorded in Romania only after approval by the Romanian Automotive Register, RAR (Art. 1(1)). The Ministry of Transport notifies RAR to the European Commission as the type-approval authority and technical service (Art. 7^2).
- Regulation (EU) 2018/858 allows recognition. Article 39(5) says approval authorities of other member states may accept the provisional approval, notifying the issuing authority in writing. It is not an obligation.
- The Road Code bans letting go of the wheel. Art. 99(1) point 13 of Emergency Ordinance (OUG) 195/2002 imposes a class I fine for "leaving free, while driving, the steering wheel, handlebar or control lever of the vehicle". According to RDW, the European version of FSD Supervised lets the driver keep their hands off the wheel, provided they can take over immediately.
- RAR's position, as reported. On 14 June, the site Not an FSD Tracker published a reply attributed to RAR's technical director, Cristian Bucur. According to that report, RAR does not recognise the Dutch provisional approval, is analysing the RDW file and cites Article 99 of the Road Code. The same report says RAR flagged to the Ministry of Transport and the Ministry of Internal Affairs, including in the Interministerial Council for Road Safety, a possible mismatch between EU type-approval law and national traffic law. We could not find this letter published on RAR's website or confirmed by other outlets, so we treat it as a report, not as a verified official position.
Our analysis: three changes, in logical order
- An administrative decision by RAR (recognise or not). Going by the text of Article 39(5), recognition looks like a decision for the approval authority, not a new law. That is how the agencies in Slovenia, Croatia and Estonia proceeded. RAR can take that decision, refuse it or wait for the EU vote.
- An amendment to the Road Code. Even with the approval recognised, Art. 99(1) point 13 remains a problem for a system that allows hands-off driving. One possible fix would be an explicit exception for assistance systems approved under EU law or UN regulations, while clearly keeping the driver responsible. The implementing regulation of the Road Code would need to be updated accordingly. The initiative can come from the Government, usually through the Ministry of Internal Affairs, which is in charge of traffic rules, or from Parliament.
- The EU vote. If the Commission authorises the exemption in December, the approval becomes valid across the EU, and RAR's step in point 1 is no longer needed. Our interpretation is that the Road Code issue would remain: a type-approval says the car complies, it doesn't change the traffic rules. It's also worth watching how Romania's representative votes in the TCMV, because the votes of mid-sized countries, not just the big ones, count towards the qualified majority.
What shouldn't change is responsibility. Every authority that has approved the system, from RDW to Slovenia's AVP and the Czech ministry, insists that the driver remains responsible for driving.
The Netherlands: how it copes with cyclists
Dutch traffic is probably Europe's toughest test: bike lanes everywhere, trams, and fast e-bikes and "fatbikes". The reports are mixed, and we present them as they are.
- Reuters (20 April) rode with a Tesla driver, Kees Roelandschap, through the narrow canal-side streets of Amsterdam. He said he had yet to run into a difficult situation among cyclists coming from every direction. On the other hand, Esther van Garderen, director of the Fietsersbond (the Dutch cyclists' union), said her organisation is not fully convinced that these cars understand how cyclists behave. Amsterdam's traffic alderman, Melanie van der Horst, said the city had no say in the approval and will monitor the rollout closely. She added, though, that having a supervising driver means "we don't need to panic".
- Flemish minister De Ridder, in her approval letter for Belgium, wrote (according to EVwire) that testing showed defensive, cautious and anticipatory behaviour that takes pedestrians, cyclists and other drivers into account.
- CleanTechnica (29 September) compared FSD with XPeng's system in a test arranged by XPeng, using a customer's Tesla Model 3. The author describes a "react and correct" style. FSD braked abruptly for cyclists still far away and left large gaps, which may be safer for them, but once it stopped in a bike lane and blocked a cyclist coming the other way. At other times it hesitated at junctions and other drivers honked. Given who arranged the test, we treat it with caution.
- Creator Steven Peeters, together with Kees Roelandschap, published an uncut drive through central Amsterdam. In his description he writes that they met plenty of cyclists and pedestrians, had "zero issues" with FSD, and needed one intervention, caused by the road layout.
The overall picture: FSD appears to see and respect cyclists, but it's often overcautious or clumsy in dense traffic. That's exactly why it remains "supervised".
What FSD looks like in the newest countries
We looked for recent clips filmed after the approvals in Slovenia, Czechia and Croatia. The descriptions below are the creators' own, not our conclusions.
Slovenia. Jan Skočaj filmed his first FSD drive in Slovenia on the evening he received the update (version 14.2.2.6), on 9 September. Avtomobilnost, a show on Slovenian public television, tested the system in Ljubljana: roundabouts, traffic lights, narrow streets with cyclists, and how strictly it monitors driver attention.
Czechia. Creator robot*irl filmed his first FSD drive through Prague in a Model Y on HW4. The Czech show STANDASHOW took the car on a "stress test" through Anděl and Vinohrady, among trams and cyclists. The host says he touched the wheel only once in 700 km.
Croatia. In a clip published on 1 October, two days after approval, Jan Skočaj drove through Istria from Novigrad to the Plovanija border crossing. He describes a drive with no takeovers, but with a few manual speed adjustments where speed-limit signs were missed, and the first serious potholes he'd met in more than 3,500 km of FSD driving in Slovenia.
For context on European alternatives to FSD, see our article on the Volkswagen–Wayve partnership.
A call to Romanian journalists and creators
This article is a starting point, not the final word. Eight European countries, some with roads and budgets similar to ours, have already made a decision. In Romania, the topic is discussed almost only in Tesla owners' groups.
If you're a journalist, content creator or podcaster, or simply care about road safety, we invite you to pick up the topic. Use the facts here, with attribution or a link to this article. Above all, put the questions that matter to the people who can answer them:
- RAR: has it analysed the RDW file, and what did it conclude?
- The Ministry of Transport and the Ministry of Internal Affairs: is there a plan for Article 99 of the Road Code?
- How will Romania vote in the TCMV in December?
We're not asking you to agree with us. We're asking that together we turn this into an informed public discussion, with the data on the table. If you write about it, let us know and we'll gladly quote you.
What it means for drivers
- In Romania, FSD (Supervised) is not approved and must not be used on public roads. The Dutch approval isn't recognised, and the Road Code penalises letting go of the wheel.
- The EU vote is scheduled for December 2026 at the earliest. Even with a "yes", a clarification in the Road Code will probably be needed.
- Where it is approved, the driver remains responsible. The system monitors attention and temporarily disables itself if ignored.
- Tesla's safety data comes from the manufacturer. The strongest independent assessment available is RDW's, after more than 18 months of testing.
- If you want the topic to move forward, questions to RAR, the Ministry of Transport and the Ministry of Internal Affairs are the most direct route.
