13.07
per 100,000 people, 2023
Moderate trust, 61 of 100 The sources disagree: their figures run from 11.93 to 18.38.
- WHO register
- WHO estimate
- ITF
Recorded change in how a source reports; hover or focus the ring for details.
Show the data as a table
| Year | WHO Mortality Database, road traffic deaths | WHO Global Health Estimates 2023, road injury | OECD/ITF, annual road fatalities |
|---|---|---|---|
| 1980 | 22.86 | – | 22.23 |
| 1981 | 21.89 | – | 21.23 |
| 1982 | 19.31 | – | 18.73 |
| 1983 | 18.56 | – | 17.97 |
| 1984 | 19.12 | – | 18.49 |
| 1985 | 18.77 | – | 18.13 |
| 1986 | 19.44 | – | 18.89 |
| 1987 | 19.43 | – | 18.85 |
| 1988 | 19.54 | – | 18.96 |
| 1989 | 18.77 | – | 18.18 |
| 1990 | 18.43 | – | 17.6 |
| 1991 | 16.9 | – | 16.18 |
| 1992 | 15.68 | – | 15.12 |
| 1993 | 15.87 | – | 15.29 |
| 1994 | 15.94 | – | 15.34 |
| 1995 | 16.11 | – | 15.59 |
| 1996 | 16.03 | – | 15.53 |
| 1997 | 15.82 | – | 15.36 |
| 1998 | 15.61 | – | 15.03 |
| 1999 | 15.53 | – | 14.96 |
| 2000 | 15.26 | 15.62 | 14.9 |
| 2001 | 15.27 | 15.64 | 14.84 |
| 2002 | 15.64 | 16.11 | 14.98 |
| 2003 | 15.38 | 15.85 | 14.79 |
| 2004 | 15.23 | 15.65 | 14.63 |
| 2005 | 15.2 | 15.56 | 14.71 |
| 2006 | 14.96 | 15.42 | 14.3 |
| 2007 | 14.32 | 14.73 | 13.67 |
| 2008 | 12.87 | 13.26 | 12.27 |
| 2009 | 11.61 | 11.96 | 11 |
| 2010 | 11.2 | 11.55 | 10.61 |
| 2011 | 11.1 | 11.43 | 10.34 |
| 2012 | 11.39 | 11.73 | 10.65 |
| 2013 | 10.92 | 11.25 | 10.28 |
| 2014 | 10.8 | 11.12 | 10.13 |
| 2015 | 11.47 | 11.83 | 10.88 |
| 2016 | 12.15 | 12.56 | 11.48 |
| 2017 | 12.05 | 12.44 | 11.28 |
| 2018 | 11.69 | 12.1 | 10.99 |
| 2019 | 11.51 | 11.92 | 10.76 |
| 2020 | 12.37 | 12.75 | 11.49 |
| 2021 | 13.75 | 14.16 | 12.62 |
| 2022 | 13.52 | 16.17 | 12.45 |
| 2023 | 13.07 | 18.38 | 11.93 |
All countriesAll of United States's figuresCompare
United States among 204 countries
79 have a higher figure and 124 a lower one. Definitions differ between countries, and their years run from 2010 to 2025.
What this number counts
- Counts
- Deaths from road traffic crashes per 100,000 people, counted under each source's own rules
- Where
- United States of America
- When
- 2023 (one year, not an average)
Where the number comes from (3 sources)
| Source | Figure and year | Definition | Quality | Counts for |
|---|---|---|---|---|
WHO Mortality Database, road traffic deathsWorld Health Organization (WHO) · Death and health registersRetrieved 2026-10-25 | 13.072023 | Registered deaths whose underlying cause is a road traffic injury, all ages and both sexes, divided by UN World Population Prospects 2024 population for 1 July, because WHO's population file has no figure for this year.On WHO's road injury list (GHE 2023 methods paper, Annex Table A, cause 1530, note j). Countries reporting 4-character ICD-10 codes (lists 104, 10M): V01-V04 and V06 .1-.9; V09.2-.3; V10-V14 and V20-V28 .3-.9; V15-V19 and V29-V79 .4-.9; V80.3-.5; V81.1; V82.1, .8-.9; V83-V86 .0-.3; V87.0-.9; V89.2-.3, .9; V99; X59.4; Y32; Y85. Countries reporting 3-character codes (list 103): V01-V04, V06, V09-V80, V87, V89, V99, Y32, Y85, which also take in some codes the 4-character list leaves out. Codes in this year's file: ICD-10 4-character codes. No 30-day limit: a death counts when the certificate gives this cause. National rows only. Year rule: the WHO register, Eurostat and ITF lines choose first. Those with data from 2021 on take the latest such year they all have, and a line that ends earlier keeps its own latest year; with no shared year from 2021 on, each keeps its own latest. WHO's estimate is set afterwards: if those lines all landed on one year from 2021 on it takes that year, capped at 2023 where it ends, and otherwise it takes 2023. So a row can show different years even where its series share an earlier year. This row: WHO register 2023, WHO estimate 2023, ITF 2023. The yearly series also covers years before ICD-10, which count ICD-9 Basic Tabulation List item B471 (E810-E819, motor vehicle traffic accidents), with population from the same sources. Raw files: morticd10_part1-6.zip (44901 deaths, of which 1155 are code Y85 and 3 are code Y32, road traffic injuries of unknown intent) and WPP2024_Demographic_Indicators_Medium.csv.gz (https://population.un.org/wpp/). WHO publishes the data as the country sent them and does not adjust for incomplete registration. | 4 of 5 checks
| 33% |
WHO Global Health Estimates 2023, road injuryWorld Health Organization (WHO) · Model-based estimateRetrieved 2026-11-02 | 18.382023 | WHO's modelled estimate of deaths from road injury (cause 1530), all ages and both sexes, per 100,000 people.For 2000-2021 WHO uses the estimates prepared for its Global status report on road safety 2023, built from death registers and from road deaths reported by governments, with a regression model for countries without eligible registers. WHO extrapolated 2022 and 2023 from them. No 30-day limit. Year rule: the WHO register, Eurostat and ITF lines choose first. Those with data from 2021 on take the latest such year they all have, and a line that ends earlier keeps its own latest year; with no shared year from 2021 on, each keeps its own latest. WHO's estimate is set afterwards: if those lines all landed on one year from 2021 on it takes that year, capped at 2023 where it ends, and otherwise it takes 2023. So a row can show different years even where its series share an earlier year. This row: WHO register 2023, WHO estimate 2023, ITF 2023. Raw file: ghe_full_road_injury.json (WHO API, GHE_FULL table). WHO gives 63132 deaths for this year, with an uncertainty range of 56274 to 69726. | 3 of 5 checks
| 25% |
OECD/ITF, annual road fatalitiesOrganisation for Economic Co-operation and Development (OECD) · Administrative recordsRetrieved 2026-11-02 · Data (API) | 11.932023 | People killed in road crashes within 30 days of the crash, as ITF's dataset describes them.Divided by UN World Population Prospects 2024 population for 1 July (estimates to 2023, medium projection after), so the whole series has one denominator. ITF's dataset does not say which authority collects the figure, so it is typed as an administrative record. Year rule: the WHO register, Eurostat and ITF lines choose first. Those with data from 2021 on take the latest such year they all have, and a line that ends earlier keeps its own latest year; with no shared year from 2021 on, each keeps its own latest. WHO's estimate is set afterwards: if those lines all landed on one year from 2021 on it takes that year, capped at 2023 where it ends, and otherwise it takes 2023. So a row can show different years even where its series share an earlier year. This row: WHO register 2023, WHO estimate 2023, ITF 2023. Raw files: trends_safety.csv (40990 deaths) and WPP2024_Demographic_Indicators_Medium.csv.gz (https://population.un.org/wpp/). ITF's own published rate for this year is 12.17. | 5 of 5 checks
| 42% |
- WHO register33%
- WHO estimate25%
- ITF42%
How the trust score was worked out
- Quality33.3 / 40
How well the sources are made
- Agreement0.4 / 30
How close the sources are to each other
- Independence20 / 20
How many separate ways the number was measured
- Freshness7 / 10
How recent the newest data is
Who reports it, and since when
No source of these kinds: police and justice records.
- WHO register, 1999: WHO's data for this country switch from ICD-9 to ICD-10 in 1999: road deaths are WHO's road injury list in ICD-10 codes from 1999, and ICD-9 Basic Tabulation List item B471 (E810-E819, motor vehicle traffic accidents) up to 1998. source
- WHO estimate, 2022: From 2022 WHO's estimates are extrapolated from the 2021 estimates of its road safety report, using available death registers and cause fractions from GBD 2023. source
figure no figure here, inside its range no later figure recorded change
Why the numbers differ
- The 30-day limit. WHO says its estimates follow ICD-10 and count all deaths in a calendar year that result from a road crash, whatever the time since the crash, unlike many official road traffic records, which use a 30-day definition. (source, retrieved 2026-11-02)
- The definition Eurostat and ITF share: a person killed is anyone dying immediately or within 30 days as a result of an injury accident, excluding suicides. Countries that do not apply the 30-day limit get a conversion coefficient. (source, retrieved 2026-11-02)
- Death certificates and road crash figures. WHO's register line counts cause-of-death codes sent by national statistics offices or health ministries. Eurostat and ITF count deaths in road crashes within 30 days. In the 46 countries with both for the same year from 2018 on, the register rate is the higher one in 17, and the median register rate is 0.96 times the Eurostat or ITF rate. (source, retrieved 2026-10-25)
- The Netherlands. CBS builds its road death count from death certificates, files from public prosecutors and Rijkswaterstaat's crash register. ITF's count equals it in 30 of 30 years, so that line shares death certificates with the WHO register line. The values still differ, and both are counted, so they are not two fully independent sources. (source, retrieved 2026-11-02)
- Eurostat says police-registered deaths in the Netherlands are under-reported, at around 84% of the national total: 566 in its CARE data against 675 published nationally in 2024. Across countries it says the data are compiled by police, ministries or statistical authorities, so only its Dutch line is typed as police records here. (source, retrieved 2026-11-02)
- Where European crash data come from. The European Commission says national data sets enter its CARE database with their national structure and definitions, and that countries have their own registration systems, not always compatible with CARE. (source, retrieved 2026-11-02)
- The Commission's common data structure for that database is also offered as a recommendation for national police road accident data collection reports. (source, retrieved 2026-11-02)
- Under-reporting. WHO reports that across countries taking part in its road safety reports, its estimates average 1.7 to 1.8 times the deaths governments report. In a record-linkage study in Lusaka Province, Zambia, it reports police records as about 19% complete, hospital records 12% and vital statistics 14%. (source, retrieved 2026-11-02)
- Modelling. WHO's estimate is not a count. Countries with death registration at least 80% complete use those data, and for countries without eligible registers WHO uses a negative binomial regression. (source, retrieved 2026-11-02)
- WHO says road injury deaths in its estimates were adjusted upwards where needed to take account of surveillance data from countries, and that the time series were extrapolated to 2023 using available death registers and cause fractions from GBD 2023. (source, retrieved 2026-11-02)
- Extrapolated years. Among the 62 countries with a usable WHO register rate for 2023, the estimate for 2023 is at least 25% above the register rate in 25, including the United States of America (18.38 against 13.07). (source, retrieved 2026-11-02)
- One set of figures, two publishers. Eurostat and ITF counts match for 31 European countries, identically in 641 of 742 country-years, so one of the two is marked as a copy there and counts once. They differ in Netherlands (Kingdom of the) (566 against 675 in 2024), so both count there. (source, retrieved 2026-11-02)
- Which codes count. The register line uses WHO's road injury list, which is finer on 4-character ICD-10 codes than on 3-character ones. Each line's definition gives the codes. Before ICD-10, the series use ICD-9 motor vehicle traffic accidents (E810-E819) and ICD-8 motor vehicle accidents (E810-E823). (source, retrieved 2026-11-02)
- Condensed lists. 6 countries' latest files have only one item for all transport accidents (V01-V99), which also counts rail, water and air deaths, so those lines carry half weight. (source, retrieved 2026-10-25)
- Population. Eurostat divides by the average of two 1 January populations and publishes one decimal. The WHO register line uses WHO's population file. The ITF line here uses UN population, so it can differ slightly from the rate ITF publishes. (source, retrieved 2026-11-02)
Common mistakes
- Comparing a rate per 100,000 people with a rate per billion vehicle-kilometres or per 10,000 road motor vehicles. ITF publishes all three, and each has a different base. (source, retrieved 2026-11-02)
- Reading a WHO estimate as a count. For countries without eligible death registers it comes from a regression model. (source, retrieved 2026-11-02)
Download, cite and report an error
From the table How many people die on the roads in each country? Updated 2026-11-02.
Download every value in this table (CSV)
Cite as: RealStats, “Road deaths per 100,000 people in United States”, updated 2026-11-02, with the original sources listed on the page.