Tuesday, 11 August 2026
Your paper, your pace.

Roads: what actually stops people dying, and what it costs

Norway lost 1.6 people per 100,000 to its roads in 2024. The United States lost 11.6. Measured per kilometre driven the gap is half that size, and the missing half is the whole argument.

Checked on

All ten problems

In 2024 Norway lost 1.6 people per 100,000 to its roads and the United States lost 11.6. Those are the counts in the International Transport Forum’s Road Safety Annual Report 2025, published on 12 December 2025, divided by population — and that fraction is the number this whole page is argued against. Sweden is at 2.0, Britain at 2.4, Germany at 3.3, the Netherlands at 3.8, Czechia at 4.5, Poland at 5.2. You can redo any of them with a calculator: Czechia had 494 deaths and 10.90 million people.

Now read the table a second time, because another one is hiding inside it. Count deaths per kilometre driven instead of per person and the order changes. Poland loses 7.3 people per billion vehicle-kilometres; the United States loses 7.8. Two countries whose per-head figures differ by more than a factor of two carry almost the same risk per kilometre. The difference is that Poles drive less. Whenever a government quotes one of these numbers, it is worth asking which of the two it picked, and why.

Roads are the big problem where the physics is least disputed and least acted on. Kinetic energy rises with the square of speed, so a body struck at 50 km/h absorbs nearly three times the energy of a body struck at 30. The power model — Rune Elvik’s 2009 report for the Norwegian institute TØI is the standard version — puts the exponent linking speed to fatalities at about 3.0 in towns and about 4.6 on rural roads. None of this is contested. What follows is five places that acted on it and measured what happened, including one country that counted 331 saved lives and then had the measure repealed underneath it, and one that wrote a rule about fuel and changed the shape of the front of every new car.

How this is measured

One number, named and sourced. Everything on this page is argued against it, so you can check us.

Road deaths per 100,000 inhabitants in 2024 — everyone killed in road traffic, counted against the population, not against the driving per 100,000 people

ITF/IRTAD Road Safety Annual Report 2025 (published 12 December 2025), Table 3 road deaths for 2024, against population at 1 January 2024. Where the report prints the rate itself — Norway 1.6, Sweden 2.0, United Kingdom 2.4, United States 11.6 — the two agree.

Who is actually ahead

Ranked on the number above — not on reputation.

Road deaths per 100,000 inhabitants in 2024 — everyone killed in road traffic, counted against the population, not against the driving
Country Result Why
Norway 1.6 87 deaths in a country of 5.55 million, and the best in the world on the other measure too: 1.9 per billion vehicle-kilometres. Norway is small enough that one bad year shows — 2025 brought 111 deaths, a rise of 28%.
Sweden 2.0 213 deaths among 10.55 million people; 2.6 per billion vehicle-kilometres. In 2025 Sweden recorded 203, the lowest count since the Second World War, against 591 in the year 2000.
United Kingdom 2.4 1,671 deaths in Great Britain. Britain has sat near the top of this table since the 1990s without a Vision Zero law, mostly through vehicle standards, drink-drive enforcement and speed cameras.
Germany 3.3 2,770 deaths among 83.5 million people. Germany is the one rich country with no general motorway speed limit, and it is mid-table rather than bottom — the autobahn is not where most Germans die.
Netherlands 3.8 675 deaths among 17.94 million. The country with Europe's most separated cycle infrastructure is going backwards: 759 died in 2025, 84 more than in 2024.
Czechia 4.5 494 deaths among 10.90 million people. Czechia is close to the EU average, which the European Commission put at 43 deaths per million in 2025 — that is 4.3 per 100,000.
Poland 5.2 1,896 deaths among 36.62 million. But 7.3 per billion vehicle-kilometres in 2023, against the American 7.8 — Poland looks twice as safe as the United States per person and almost identical per kilometre.
United States 11.6 39,345 deaths among 340 million people, and 7.8 per billion vehicle-kilometres in 2023. That is 5.8 times Sweden per person but 3.0 times Sweden per kilometre. The rest is mileage.

What has actually been tried

Real countries, real policies, and what the numbers did afterwards — including where it went wrong.

Sweden

Sweden's parliament adopted Vision Zero in October 1997: the premise that people will keep making mistakes and the road has to be built so the mistakes are survivable. The signature measure is the 2+1 road — an ordinary two-lane rural road rebuilt with alternating overtaking lanes and a steel cable down the middle, so head-on collisions become physically impossible. About 1,800 kilometres had opened by 1 January 2008.

What happened. The Swedish National Road and Transport Research Institute evaluated the whole network. Fatalities on those roads fell 76%, and 79% on the road links themselves; fatal and serious injuries together fell 57% to 63% depending on the speed limit. Nationally, road deaths went from 591 in 2000 to 203 in 2025.

The catch. Barriers separate cars from cars. They do nothing for the 95 people out of 203 who died in 2025 outside a vehicle: 28 motorcyclists, 27 pedestrians, 25 cyclists. And 39% of the dead were 65 or over, an age at which the same impact is far more likely to be fatal. Note also that the two Swedish agencies disagree — the transport agency counted 208 in a January 2026 release, the statistics office 203 in April 2026.

Carlsson, 'Evaluation of 2+1-roads with cable barrier: Final report', VTI rapport 636A, 2009; Trafikanalys, Vägtrafikskador 2025, published 27 April 2026

Netherlands

The Netherlands did not mainly lower speed limits on paper; it rebuilt streets so the lower speed felt natural — narrowing, raised junctions, continuous footways, separated cycle tracks. By 2007 roughly 75% of Dutch residential streets sat inside 30 km/h zones. On 8 December 2023 Amsterdam extended 30 km/h to most of its main city streets as well.

What happened. The standard evaluations put the effect of converting a residential area from 50 to 30 km/h at a 22% to 25% fall in injury crashes. Amsterdam's own one-year assessment, published 5 June 2025, found motor vehicle crashes down from 926 to 823, a fall of 11%, and injuries down from 275 to 252. Deaths were unchanged at 4.

The catch. It has stopped working nationally. Dutch road deaths rose from 675 in 2024 to 759 in 2025. Of those, 281 were cyclists, at least 41% of them on e-bikes, and 118 were men aged 70 or over — 40 more than the year before. Street design cannot stop an 80-year-old falling. Note also that the European Commission published a 2025 rate of 35 per million for the Netherlands using only the first six months; the full Dutch year works out at about 42.

SWOV fact sheet 'Zones 30: urban residential areas' (January 2009), citing Elvik 2001 and Vis & Kaal 1993; City of Amsterdam, '30 km/h in Amsterdam' research report, 5 June 2025; CBS, 9 April 2026

United Kingdom

On 17 September 2023 Wales became the first country in Europe to make 20 mph — about 32 km/h — the default limit on restricted roads. That meant almost every residential street changed at once, unless a council actively decided otherwise. It was a change to the default, not a building programme: new signs and a new starting assumption.

What happened. Transport for Wales measured mean speeds falling from 28.7 mph to 24.9 mph. In the first twelve months, casualties on 20 and 30 mph roads fell from 2,257 to 1,772, a drop of 26.2%, while casualties on roads limited to 40 mph and above rose 4.2%, from 2,164 to 2,257. Casualties across Wales fell 11.8%.

The catch. The politics cost more than the policy. A petition to repeal the order collected 469,571 signatures, the largest in the Senedd's history; the government reopened its guidance on exceptions and councils began restoring 30 mph on selected roads. Transport for Wales also warns that three years of data are needed before pedestrian and cyclist deaths can be read at all, because the annual counts are too small to be stable.

Transport for Wales, 'Default 20mph speed limit on restricted roads: national monitoring report', July 2025; Senedd Research

France

On 1 July 2018 France cut the limit on rural two-way roads without a central barrier from 90 to 80 km/h. Those roads carry the majority of French road deaths. The measure cost almost nothing to introduce: new signs and the cameras that were already there.

What happened. Cerema's evaluation, published July 2020, measured average speeds falling 3.3 km/h, from 86.4 to 83.1. It counted 331 fewer deaths on the affected network over the first 18 months and estimated 349 over 20 months, measured against a 2013–2017 baseline with the unaffected road network as a control.

The catch. This is the attempt that was taken away. Speed enforcement became the symbol of a distant state taxing rural drivers, and by 8 January 2019 the interior minister said nearly 60% of France's 3,200 speed cameras had been neutralised, attacked or destroyed. A 2019 law let departmental councils restore 90 km/h; from January 2020 they began doing so, Haute-Marne first. The lives were counted and the mandate was gone.

Cerema for the Interministerial Delegation for Road Safety, evaluation of the 80 km/h limit, July 2020; BBC News, 8 January 2019

United States

The United States regulated the inside of the car and left the outside alone. From 2011 its fuel economy rules set each model's target by its footprint — track width times wheelbase — so a physically bigger vehicle is held to a weaker standard. Whitefoot and Skerlos modelled that incentive in Energy Policy in 2011 and found it could raise the sales-weighted average vehicle size by 1 to 16 square feet and cancel 1 to 4 mpg of the intended saving.

What happened. The fleet grew tall. The Insurance Institute for Highway Safety analysed 17,897 crashes involving one vehicle and one pedestrian and reported in November 2023 that vehicles with a leading hood edge above 40 inches are about 45% more likely to kill the pedestrian they strike than low, sloping-fronted cars; blunt fronts of medium height add about 26%. US pedestrian deaths rose from 4,109 in 2009 to 7,080 in 2024, up 72% from the low point, while total road deaths fell to an estimated 36,640 in 2025.

The catch. Vehicle shape is not the whole story. Tyndall estimated in 2021 that if the growth in SUVs had been cars instead, about 1,100 pedestrian deaths would have been averted between 2000 and 2019 — real, but a fraction of the rise. Phones, higher speeds, more walking after dark and roads built with no footway are all in the same data. The claim here is narrow: a rule written about fuel changed the shape of the front of a car, and that shape is measurable in the mortality figures.

Whitefoot and Skerlos, Energy Policy, 2011; IIHS (Hu and colleagues), 14 November 2023; IIHS Fatality Facts 2024: Pedestrians, from FARS, posted June 2026; NHTSA, 1 April 2026

What a machine would optimise for

This is arithmetic, not advice and not a prediction. We name the single number being maximised and follow it wherever it goes. The point is to see the shape of the answer a calculator gives.

The number being maximised

One number only: road deaths per 100,000 inhabitants. Journey time, freight costs, rural access and personal liberty are absent from the calculation, because none of them appears anywhere in that fraction.

  1. Buy the largest measured reduction per euro first, and that is a steel cable. Sweden's 2+1 conversions cut fatalities on those roads by 76%, because an unbarriered rural road is the one place where an ordinary lapse of attention becomes a head-on collision at a closing speed near 160 km/h.
  2. Set 30 km/h wherever people walk, and pay for it in seconds. Amsterdam measured the entire trade in one report: crashes down 11% in a year, a 30-kilometre car journey longer by about 30 seconds, a ten-minute bus ride longer by 13.
  3. Regulate the front of the vehicle, not only the cabin. A leading hood edge above 40 inches raises the chance of killing the pedestrian it hits by roughly 45%, and that is a geometry a rule can specify, exactly as rules already specify crumple zones and airbags.
  4. Read the denominator. It is people, not kilometres, so a kilometre never driven scores identically to a kilometre made safe. Poland is at 5.2 per 100,000 and the United States at 11.6, yet per billion vehicle-kilometres they are 7.3 and 7.8: the same road risk, half the death rate, because Poles drive less.

The arithmetic

United States 2024: 39,345 deaths, 11.6 per 100,000. Sweden 2024: 213 deaths, 2.0 per 100,000. The ratio is 5.8. Per billion vehicle-kilometres the same two countries are 7.8 (US, 2023) and 2.6 (Sweden, 2024) — a ratio of 3.0. Divide one ratio by the other: 5.8 / 3.0 = 1.9, and that leftover factor is mileage, since an American drives roughly twice as far in a year as a Swede. The sum comes out like this. Keep every American kilometre but make each one as safe as a Swedish kilometre and the annual toll is 39,345 / 3.0 = about 13,100, some 26,200 fewer than today. Apply Sweden's rate to 340 million Americans and it is about 6,800. The last 6,300 people are not on the road at all. They are in the odometer.

What the machine would miss

The most important part of this page. A number that goes up can still be paid for by somebody, and some things never make it into the number at all.

Who pays

Drivers pay in seconds, every single day, for a crash that then does not happen to them. Amsterdam priced it honestly: 30 seconds on a 30-kilometre car journey, 80 seconds on a full bus route. In Wales the mean speed fell 3.8 mph and 469,571 people signed to repeal the law, the largest petition the Senedd has ever received. In rural France the bill was heavier because there is no bus, and by January 2019 nearly 60% of 3,200 speed cameras had been attacked or destroyed.

What the number flattens

Deaths per head rewards countries where people drive less, and cannot tell that apart from countries where driving is safe. Poland sits at 5.2 per 100,000 and the United States at 11.6, yet Poland loses 7.3 people per billion vehicle-kilometres and America 7.8: near-identical risk on the road, less than half the deaths per person, because Poles cover fewer kilometres. The measure also counts only people who were out there. It has no column for the child driven to school because the crossing frightened her mother.

What a spreadsheet cannot see

In 2025, 118 Dutch men aged 70 or over died on bicycles, 40 more than the year before, most of them on e-bikes. In the table each one is a failure of street design. In life the e-bike is the reason a man of 78 still does his own shopping, sees people and does not spend the winter indoors. The alternative to a risky trip is very often not a safer trip but no trip, and no road safety statistic in any country records the journeys that were never made.

Where does your country stand?

Open my country

Where the numbers come from

We are a newspaper, not a government and not an adviser. Nothing on this page is a recommendation to you or to anyone in office. It is what was tried, what it measured, and what a calculator would say if you let it loose on the same problem.

One good piece of thinking a day

The day's most worthwhile story, and the question underneath it. No spam, one click to leave.

One click to leave. We never sell or share your address.