Europe is removing e-scooters city by city. Nobody checked whether the riders were sober.
On 1 September, thousands of shared scooters and bikes come off the streets of Brussels. Paris removed its rental fleet in 2023. Madrid revoked operator licences in 2024. Prague pulled the plug in January.
Four European capitals in three years. So it’s worth reading what they said the problem was. Scooters on pavements. Scooters dumped across footpaths. Parking chaos. Rising injury counts.
Every one of those is real. And every one of them describes where the scooters ended up, not why the rider went down.
Brussels counted 666 injured e-scooter riders in 2025, up 26% in a year, then made a decision affecting over a million users. Nowhere in the public record is there a figure for how many of those 666 had been drinking.
Not because nobody wondered. Because nobody measured.
That gap is why we’re writing this. In July, researchers at Chalmers University of Technology and the Swedish Transport Administration published the number Europe had been arguing without.
The finding
The median person killed on an e-scooter in Sweden had 1.8‰ alcohol in their blood.
The legal limit is 0.2‰. Nine times over.
Not one of them was wearing a helmet.1
One number to remember
44% of e-scooter riders killed in Sweden had been drinking.
Among cyclists killed in the same database: 12.5%. Among e-bike riders: 26.5%.1
Same country. Same streets. Same database. Alcohol turns up three and a half times more often on an e-scooter than on a bicycle.
Not a handful of cases. Nine thousand.
Sweden’s national transport research institute went through every recorded e-scooter injury crash from 2019 to late 2023- 9,271 of them, drawn from police and hospital records. Ten dead. 322 seriously injured. 2,744 moderately injured.8
Look at when they happen. Almost half of all e-scooter accidents in Sweden strike at night – on a vehicle sold as a way to get to work.9
And when Stockholm’s emergency department checked who these riders actually were, the pattern was stark: 63% of those wheeled in between 22:00 and 06:00 had been drinking. In daylight hours, it was 22%.2 After dark, most of the people crashing a rented scooter are drunk.
Every city that looks, finds it
One study in one country proves nothing. So we went looking. Every European study that has tested e-scooter riders – in hospital or at the roadside – has found the same thing.
- Leipzig: 53.5% of 411 police-recorded e-scooter cases were alcohol-positive, at a mean of 1.42‰.10
- Stockholm: 42.9% tested positive for alcohol overall.2
- Hamburg: 33% arrived under the influence, at a median of 1.5‰. One patient out of 268 was wearing a helmet.3
- Munich: 30%, at an average of 2.1‰. Of those, 78% came in between 22:00 and 06:00.4
- Helsinki: 29%. Cyclists in the same city, same period: 8%. Per trip taken, scooter riders were 3.6 times more likely to end up in hospital.5
Between a quarter and well over half. Every time. In every place that bothered to test.
Alcohol doesn’t just cause the crash. It changes the injury.
This rarely makes it into a news report, and it’s the part that matters most.
- Hamburg: 84% of intoxicated riders hit their head or face. Sober riders: 46%.3
- Helsinki: head injuries in 76% of intoxicated riders.5
- Stockholm: face and head injuries in 77% of alcohol cases, against 57% without.2
And across Swedish e-scooter accidents as a whole: 44% produce a head injury. Only 13% of the riders were wearing a helmet.9
Why? A sober rider going over the handlebars gets an arm out. It’s automatic, you don’t decide to do it. But that reflex is one of the first things alcohol takes away, so they land on their face.
So why has nobody seen these numbers?
American national injury data records alcohol in 9% of e-scooter emergency visits, a fifth of what European hospitals find.6
The difference isn’t American sobriety. The US figure is whatever a clinician typed into a chart at four in the morning. The European figures came from studies that tested the patient.
If you don’t test, you don’t find it. And what you don’t find never reaches the statistics that policy gets built on.
So cities count the scooters on the pavement, because scooters on the pavement are countable and end up removing a vehicle to solve a problem that was in the rider.
Eighteen people a day
In 2025 Sweden recorded 6,624 people injured on e-scooters. That’s eighteen a day, every day, in a country of ten million.
Accidents rose 38% in a single year. Nearly half of the seriously injured were under 25. E-scooter riders now make up more than a quarter of all injured unprotected road users in the country.7
Seven people were killed. For scale: across the entire nine years to 2024, Sweden recorded eighteen e-scooter deaths in total.1
The Swedish Transport Agency’s own list of what drives serious injury runs to three items. The first is alcohol.
The moment nobody uses
There are a few seconds between someone lifting their phone at a scooter and the wheels turning. They’re standing still, holding a device with a camera in it, and nothing has happened yet.
That is the last moment in the entire sequence when anything can be established about the state they’re in. After it there’s only a police stop that almost certainly won’t happen, or an emergency department at half past two, measurement arriving far too late to be any use.
The researchers got here before we did. Having found that 44% of the dead had been drinking, the Chalmers and Trafikverket team set out what should follow, and alongside helmets and training, they called for real-time impairment detection.1
The Munich trauma surgeons, working separately, asked for the same: tighter alcohol control, and locking the vehicles during the hours when the injuries happen.4
This is what we build, and Chalmers is one of our research partners. But we didn’t put the recommendation in their papers. They wrote it because they had spent years looking at what comes through the door at two in the morning.
Brussels removes thousands of scooters on 1 September. The riders who were drinking will still be drinking.
Everyone else loses a cheap way to get across a city without a car, without emissions. Over a million people used these scooters last year and most of them sober, every time. That’s the price of banning a vehicle instead of checking who’s riding it.
The evidence is in. The counting isn’t.
Sources
- Pai, R. R., Dozza, M. & Fredriksson, R. – Journal of Safety Research, July 2026. Chalmers University of Technology and the Swedish Transport Administration; Swedish national fatality database, 2016–2024, 204 deaths. Summary
- Alcohol and electric scooter injuries in an emergency department: a prospective observational study – Scandinavian Journal of Trauma, Resuscitation and Emergency Medicine, 2025. Stockholm.
- Risk factors and injury patterns of e-scooter associated injuries in Germany – Scientific Reports, 2022. Hamburg, 268 patients.
- E-scooter accidents – epidemiology and injury patterns: 3-year results from a level 1 trauma center in Germany – Archives of Orthopaedic and Trauma Surgery, 2024. Munich, 271 patients.
- Comparing the characteristics of electric scooter and bicycle injuries: a retrospective cohort study – Scientific Reports, 2025. Helsinki, 677 e-scooter and 1,889 cyclist cases, 2022–2023.
- Alcohol-related injuries from e-scooter and e-bike use in the US (2019–2022) – BMJ Injury Prevention, 2024. 296,590 estimated emergency department visits.
- Kraftig ökning av antalet elsparkcykelolyckor – Transportstyrelsen, 2026.
- Alkohol orsak till många olyckor med elsparkcyklar – VTI, Swedish National Road and Transport Research Institute (Susanne Wallhagen). 9,271 Swedish e-scooter injury crashes, 2019–November 2023, from police and healthcare records. Source
- Flest olyckor med elsparkcyklar sker nattetid – Folksam, study of Swedish e-scooter accidents, 2019–June 2021. Source
- Driving under the influence of alcohol and drugs on e-scooters: a toxicological perspective of public health and road safety in Leipzig, Germany – International Journal of Legal Medicine, 2026. 411 police-recorded e-scooter cases, October 2021-June 2024.
