LIVE Global accident monitoring — updating every 15 minutes
Tool · official datasets · one definition per comparison

Compare Accident Trends by Place and Period

Quick answer: pick a dataset, up to six places and a period. The chart and the table below it use one definition at a time — road death rates for countries, recorded collisions for UK councils, fatal crashes for US states, monthly recorded crashes for cities — and show a rate only where the population behind it is known.

Comparison

Build a comparison

The interactive chart loads in a moment. The default comparison is in the table below.

Default comparison · World Bank indicator SH.STA.TRAF.P5 (WHO estimates), CC BY 4.0

Road deaths per 100,000 people, 2010–2019

Modelled WHO estimates, comparable between countries by design. A rate, so countries of different sizes can be compared directly.

Road traffic deaths per 100,000 people by year
Country2010201120122013201420152016201720182019Change
United Kingdom3.73.43.02.93.22.93.03.03.03.2-14%
United States11.711.611.911.411.312.112.912.812.612.7+9%
India17.217.417.617.016.115.615.315.215.515.6-9%
Pakistan14.714.813.813.713.513.513.513.313.113.0-12%
Germany4.85.34.74.44.34.44.04.04.13.8-21%
Brazil24.324.524.823.123.620.819.918.717.116.0-34%
Definitions

What each dataset measures

DatasetMeasuresSourceRate or count
Countries — road deaths per 100,000 people (190 places)Road deaths per 100,000 peopleWorld Bank indicator SH.STA.TRAF.P5 (WHO estimates), CC BY 4.0Modelled WHO estimates, comparable between countries by design. A rate, so countries of different sizes can be compared directly.
Great Britain local authorities — recorded injury collisions (350 places)Recorded collisions; Fatal and serious collisions; Casualties; People killedDepartment for Transport STATS19, Open Government Licence v3.0Counts, not rates: population and traffic are not in the data, so a large council records more. Compare trends, or use the indexed view.
US states — fatal crashes and deaths (52 places)Deaths; Fatal crashes; Pedestrians killed; Motorcyclists killedNHTSA Fatality Analysis Reporting System (FARS), public domainCounts of crashes in which someone died within 30 days. No yearly population in the data, so no rate is shown.
Cities — recorded crashes per month (official portals) (34 places)Recorded crashes per monthEach city's official open-data portal (see its city page)Every city records crashes differently (all crashes, injury crashes only, police-attended…), so compare the shape of the trend with the indexed view rather than the levels. The newest month may still be filling in.

Related: historical accident hotspots · road accident histories · city crash maps · road death statistics · US road deaths by year · methodology

Questions

Accident trends — FAQ

Why can I only compare places inside one dataset?
Because the datasets measure different things: modelled death rates for countries, police-reported injury collisions for Great Britain, fatal crashes for US states, and whatever each city portal records for cities. Putting a Great Britain collision count next to a US fatal-crash count would compare two different definitions, so the tool keeps them apart.
When does the tool show a rate?
Only when the dataset itself is a rate with a known denominator — the country series is deaths per 100,000 people. For councils, states and cities the data has no matching yearly population or traffic figure, so the tool shows counts and never divides by an estimate of its own.
What does the indexed view mean?
Each place starts at 100 in the first period you choose, and later periods show the percentage of that starting value. It lets you compare the direction and size of a change between places of very different sizes without pretending the counts are rates.
How current is the data?
Country rates run to 2019 (World Bank publishes WHO estimates with a lag). Great Britain collisions cover 2021–2025 and US fatal crashes 2020–2024, both annual official files. City series are the last 12 months from each portal and the newest month may still be incomplete.