Evidence

The problem

Europe's roads are getting safer too slowly.

Road deaths in the EU against the 2030 promise

Road deaths (EU-27)Steady path to the 2030 target

View as table

Road deaths in the EU against the 2030 promise
Year	Road deaths	Steady path
2019	≈22,800	22,800
2020	–	21,400
2021	–	20,100
2022	–	18,900
2023	–	17,700
2024	–	16,700
2025	19,400	15,600
2026	–	14,700
2027	–	13,800
2028	–	12,900
2029	–	12,200
2030	–	11,400

Dashed line: a steady path from 2019 to the 2030 target, falling 6.1% a year. Delta-V calculation from European Commission figures1. 2019 is derived from the Commission's table, which puts 2025 15% below 2019, rounded.

19,400

people killed on EU roads in 2025, preliminary figure

SourceEuropean Commission, “EU road deaths drop by 3% in 2025”, 24 March 2026

Link to this figure

≈22,800

road deaths in 2019, the baseline year of the EU target

Delta-V calculationDerived. The Commission's table puts 2025 15% below 2019, rounded, so 2019 was about 19,400 ÷ 0.85.

SourceEuropean Commission, “EU road deaths drop by 3% in 2025”, 24 March 2026

Link to this figure

−15%

change in EU road deaths from 2019 to 2025, against a promise of −50% by 2030

SourceEuropean Commission, “EU road deaths drop by 3% in 2025”, 24 March 2026

Link to this figure

100,000

people seriously injured on EU roads every year

SourceEuropean Commission, “EU road deaths drop by 3% in 2025”, 24 March 2026

Link to this figure

≈3,800

more deaths in 2025 than a steady path to the 2030 target allows

Delta-V calculationA steady path halves deaths between 2019 and 2030 by the same share each year: 6.1% fewer every year. That path reaches about 15,600 in 2025 and 11,400 in 2030. 19,400 minus 15,600 is about 3,800. If the Commission's rounded 15% is really 14.5% to 15.5%, the gap is 3,700 to 3,900.

SourceEuropean Commission, “EU road deaths drop by 3% in 2025”, 24 March 2026

Link to this figure

≈53 a day

deaths on EU roads at the 2025 rate, or one every 27 minutes

Delta-V calculation19,400 deaths ÷ 365 days ≈ 53 a day. 1,440 minutes in a day ÷ 53 ≈ 27 minutes. The counter on our home page runs at this rate from 1 January.

SourceEuropean Commission, “EU road deaths drop by 3% in 2025”, 24 March 2026

Link to this figure

The human factor

Machines don't drink, doze or scroll.

€280 bn

lost every year to road crashes, about 2% of EU GDP

SourceEuropean Commission, SWD(2019) 283, EU Road Safety Policy Framework 2021–2030

Link to this figure

Shares overlap: one crash can involve several factors.

The evidence

The safer driver is already on the road.

Fewer crashes than human drivers on the same roads

Reduction in crash rate, Waymo rider-only driving, data through June 2026

Serious injury or worse−95%
Pedestrian injury−93%
Cyclist injury−86%
Any injury−82%
Airbag deployment−82%
Motorcyclist injury−82%

View as table

Fewer crashes than human drivers on the same roads
Crash type	Reduction	Crashes avoided
Serious injury or worse	−95%	55
Pedestrian injury	−93%	86
Cyclist injury	−86%	56
Any injury	−82%	841
Airbag deployment	−82%	358
Motorcyclist injury	−82%	36

271 million miles

driven by Waymo without a human driver (437 million km), in Phoenix, San Francisco, Los Angeles, Austin and Atlanta, through June 2026

SourceWaymo, Safety Impact data hub, data through June 2026

Link to this figure

−95%

fewer crashes with serious injury or worse than human drivers on the same roads; see the chart for every crash type

SourceWaymo, Safety Impact data hub, data through June 2026

Link to this figure

56.7 million miles

analysed in a peer-reviewed study in Traffic Injury Prevention. In none of eleven crash types did the driverless system do significantly worse. All authors work for Waymo, and the data run to January 2025

SourceKusano et al., Traffic Injury Prevention, 2025, doi:10.1080/15389588.2025.2499887

Link to this figure

−92%

bodily-injury claims against human drivers, and −88% property-damage claims, over 25.3 million miles (40.7 million km) studied with Swiss Re

SourceWaymo, “New Swiss Re study”, 19 December 2024

Link to this figure

759

road deaths in the Netherlands in 2025, 84 more than in 2024; 281 of them were cyclists. From 2021 to 2025, 31% of cyclists killed did not collide with another road user or a fixed object

SourceCBS, “Meer verkeersdoden in 2025”, 9 April 2026

Link to this figure

These are one company's published figures. We want every system, from every manufacturer, measured the same way and in public.

The deployment gap

Europe is watching from the kerb.

500,000+

paid driverless trips a week in the United States (Waymo, September 2026)

SourceWaymo, “Building Waymo’s Risk and Insurance Foundation in Europe with Allianz Partners”, 15 September 2026

Link to this figure

350,000+

fully driverless rides in a peak week in China (Baidu Apollo Go, March 2026), present in 28 cities worldwide by August 2026, test sites included

SourcesBaidu, first quarter 2026 results (SEC filing), 18 May 2026; Baidu, second quarter 2026 results (SEC filing), 18 August 2026

Link to this figure

24 million km

driven in the Netherlands by nearly 40,000 Teslas with FSD (Supervised), without any relevant incidents, according to the RDW in June 2026. The system assists a driver who is monitored continuously. That distance is still too short to compare its safety with human drivers

The source's own words“nearly 40,000 Teslas equipped with FSD Supervised have driven approximately 24 million kilometres in the Netherlands since the type approval was granted, without any relevant incidents”

SourceRDW, “Explanation of the type approval of FSD Supervised”, 17 June 2026

Link to this figure

≈390 lives

lost by 2035 for each year that automated driving reaches EU roads later. A first estimate, with a range of 160 to 620

Delta-V calculationA first estimate from our paper The Cost of Waiting. It starts from 19,400 deaths in 2025. Car occupants are 44% of EU road deaths and 60% of deaths in crashes involving cars. So about 73% of deaths involve a car, and we use 70%. The share of new cars sold with automated driving follows McKinsey's worldwide scenarios. That is 4%, 12% or 20% in 2030, and 17%, 37% or 57% in 2035. Sales stay at 10.8 million cars a year and the fleet at 256 million. We assume these cars drive 50% of their kilometres automated, with 80% fewer fatal crashes on those kilometres. A roll-out that starts one year later costs about 390 lives by 2035. The slow and fast scenarios give 160 and 620. With 50% fewer fatal crashes instead of 80%, the estimate is about 250.

SourcesEuropean Commission, “EU road deaths drop by 3% in 2025”, 24 March 2026; European Road Safety Observatory, Facts and Figures: Car occupants, 2023; ACEA, Vehicles on European roads, 15 January 2026; ACEA, “New car registrations: +1.8% in 2025; battery-electric 17.4% market share”, 27 January 2026; McKinsey & Company, “Autonomous driving’s future: Convenient and connected”, 6 January 2023

Link to this figure

13.8 million

direct and indirect jobs supported by the automotive sector, 6.1% of all employment in the EU

SourceEuropean Commission, Automotive industry

Link to this figure

We show our working for every derived number.

≈22,800 road deaths in 2019, the baseline year of the EU target
≈3,800 more deaths in 2025 than a steady path to the 2030 target allows
≈53 a day deaths on EU roads at the 2025 rate, or one every 27 minutes
≈390 lives lost by 2035 for each year that automated driving reaches EU roads later. A first estimate, with a range of 160 to 620

Preliminary figures may change when final data is published. We update this page when that happens.

We round derived figures to the nearest hundred. Figures from a source appear as the source states them.

Waymo reports distances in miles. We give kilometres alongside, converted at the standard rate and rounded.

Every source is public and linked.

Footnote numbers stay the same across the site, so a citation keeps pointing at the right source.