What do India’s road-death figures reveal?

Is the recorded toll falling?

No. The Ministry of Road Transport and Highways (MoRTH) reports 177,175 deaths in 2024 in Road Accidents in India 2024, up from 172,890 in 2023. That is 4,285 more, an increase of 2.5 per cent after rounding. The police also recorded 487,707 accidents and 471,441 injured people in 2024. An accident is a crash event, while a death is a person, and a single event can kill more than one person.

The five-year line starts at 138,383 deaths in 2020. Traffic was disrupted by Covid restrictions that year, so the rise from that base is not a clean measure of worsening safety. Even the 2023–24 comparison is a count, not a death rate per journey. More traffic, longer travel, more severe crashes and changes in reporting could each move the annual total. The report gives rates relative to population and registered vehicles, but not the number of journeys or kilometres travelled needed to separate those explanations. What the data does show is that the police-recorded death count rose again.

How many people die per reported accident?

MoRTH recorded 21.6 people killed for every 100 reported road accidents in 2005. It peaked at 37.3 in 2021 and was 36.3 in 2024. The ministry calls this accident severity. It divides the number of people killed by the number of police-reported crash events, then multiplies by 100. In 2024, that meant 177,175 people killed across 487,707 reported accidents.

The 20-year line shows that the recorded death toll grew faster than the number of accidents entering this measure. It does not mean 36.3 per cent of accidents were fatal: one crash can kill more than one person. Table 1.6 separately puts the share of reported accidents that were fatal at 33.7 per cent in 2024. Neither measure is the chance of dying on a journey. Changes in whether nonfatal crashes reach the police record can alter the ratio. The 2020 and 2021 values also sit within disrupted Covid-era travel. The measure raises a serious question about the severity of reported crashes without identifying its cause.

Who is being killed on India's roads?

Among the people MoRTH recorded as killed in 2024, two-wheeler users were the largest victim group: 81,780 deaths, or 46.2 per cent of the total. The report also records 36,526 pedestrians and 4,161 bicyclists. Together, these three groups account for 122,467 deaths, or 69.1 per cent of the police-recorded toll.

This table describes the person who died. It does not say who caused the collision. The shares also cannot tell us the chance of dying on a given trip, because that would require knowing how far or how often each group travels. A high death count can come from frequent travel, greater danger per trip, or both. The report's "Other" category groups 11,890 deaths across several unlike road users, including people in animal-drawn vehicles, cycle rickshaws and hand carts. The summary does not separate them.

How do deaths differ for women and men?

MoRTH records 151,950 male and 25,225 female road deaths in 2024. Men account for 85.8 per cent of the police-recorded total. Those are victim counts: they do not tell us how many women and men travelled, how they travelled, or how far.

The split matters when reading the pedestrian figures. Women made up 20.5 per cent of pedestrians killed, even though they were 14.2 per cent of all road deaths. The walking article compares pedestrian deaths with all road deaths within each recorded sex. Pedestrians were 29.6 per cent of female road victims and 19.1 per cent of male road victims. That is a difference in the mix of people killed, not proof that walking is more dangerous for women. The report publishes only male and female categories.

How old were the people who died?

The 18–25, 25–35 and 35–45 age bands hold 117,026 deaths between them, or 66.1 per cent of the total. The report records 10,117 deaths of people younger than 18. It could not identify the age of 4,705 people, so that category remains in the chart. Dropping the unknowns quietly would make the other shares look more certain than they are.

The age distribution shows the human reach of the loss, including many people in the years when they may be studying, working or caring for family. It is still a distribution of deaths, not a comparison of risk between age groups. Population size, travel habits and road use differ across ages. A child in the table could have been a passenger, a pedestrian or a cyclist. The age share cannot be multiplied by a road-user share to estimate that overlap, because the separate national summaries do not contain that information.

Is this mainly an urban toll?

The police record says it is not. Rural areas account for 125,422 deaths in 2024, or 70.8 per cent of the national count. Urban areas account for 51,753. The rural share of recorded accidents is lower, at 62.4 per cent. This gap is consistent with more deaths per reported accident in the rural classification, but it does not explain why. Speed, distance to care, road and vehicle types, and differences in reporting all need separate evidence.

MoRTH places an old 2011 rural population share next to its 2024 crash totals. That should not be turned into a current rural-versus-urban death rate. Population is not road travel, the census year differs, and trips cross area boundaries. The rural and urban bars show where the recorded deaths were assigned. They do not show the risk faced by a rural resident or the safety of a comparable journey.

Which kinds of roads carry the toll?

National highways account for 64,772 deaths and state highways for 39,277. Together that is 104,049 people, or 58.7 per cent of the 2024 total. Other roads account for 73,126. That broad residual spans many settings, including local roads whose traffic conditions differ markedly from one another.

MoRTH notes that national highways were 2.14 per cent of road length as of March 2022. Placed beside their 36.6 per cent share of 2024 deaths, that gap invites a question about danger, but it cannot answer it. A kilometre of busy highway carries more vehicle travel than a kilometre of quiet local road, and the road-length measure is older than the death count. Ranking road types by danger needs deaths per kilometre driven, or per comparable journey. These figures locate the recorded deaths. They do not show which road design or driver behaviour produced them.

Which states account for the largest totals?

Uttar Pradesh records 24,118 road deaths, the most of any state in 2024. Tamil Nadu follows with 18,449, Maharashtra with 15,715, Madhya Pradesh with 14,791 and Karnataka with 12,390. These five states together record 85,463 deaths, 48.2 per cent of the national total. The chart deliberately shows a selected top five; the remaining deaths are spread across other states and union territories.

A state count is useful for sizing the administrative burden and deciding where closer local reporting would matter. As a safety league table it is poor. States differ in population, vehicles, distance travelled, road networks, health care and how completely police returns capture crashes. Dividing by population alone would still leave large differences in travel exposureExposureHow much people use a road or mode, such as journeys or kilometres travelled.Without it, a death count cannot tell us the risk of a comparable trip.. The 2024 national publication also notes a specific recast of West Bengal data using electronic accident-reporting aggregates. That is another reason to inspect reporting practice before reading a gap between two states as a gap in driving behaviour.

How were fatal crashes classified?

The collision table assigns 37,404 deaths to crashes recorded as a vehicle being hit from behind, 34,030 to hit-and-run cases and 28,400 to head-on collisions. These are large groups, and their names do different jobs. "Hit from back" and "head-on" describe the collision pattern. "Hit and run" describes what happened after the collision. The categories as published sum to the national total, and they are not a list of mutually comparable causes.

The hit-and-run count rose from 31,209 in 2023 to 34,030 in 2024. That change sits in the police classification. The table does not say whether the underlying behaviour changed, whether identification improved, or both. A rear impact likewise does not identify which driver made an error. Speed, road layout, visibility and enforcement may all matter, and the national row cannot separate their contributions. Treat this chart as a map of recorded circumstances and a starting point for investigation. It cannot deliver a causal verdict.

What does the report say about helmets and seatbelts?

MoRTH records 54,122 people killed while not wearing a helmet in 2024: 39,470 of them drivers and 14,652 passengers. It records 14,466 deaths involving non-use of a seatbelt, split between 7,489 drivers and 6,977 passengers. These are concrete counts of what police reports recorded about protective-device use among the people killed.

These counts do not estimate how many deaths a helmet or seatbelt would have prevented. We do not know what would have happened in each crash if the victim had worn one. The table also does not count all riders and occupants by whether they used a protective device, and the helmet and belt groups refer to different vehicle situations. Dividing either death count by all road deaths would not measure the risk of riding without protection. The police record identifies a serious problem; separate studies are needed to measure effectiveness and compliance.

When are crashes recorded during the day?

The largest three-hour window is 6 pm to 9 pm: police recorded 102,897 accidents then, 21.1 per cent of the 2024 total. The next largest is 3 pm to 6 pm, with 85,010. Time was unknown for 6,118 accidents, which stay in the chart rather than being assigned to a convenient hour.

These are accident counts. They do not count deaths, and the report does not say that 6 pm to 9 pm was the deadliest period for people killed. The table also does not measure vehicles on the road by hour. A busy interval could record many crashes even if the chance of a crash per journey were unchanged. Daylight, commuter traffic and road conditions may be relevant, and this national time table cannot divide their effects. It does tell an investigator where to ask for finer local crash, traffic and lighting data.

How complete is the police record?

The World Health Organization (WHO) offers an independent check for 2021, not 2024. Its India country profile reports 153,972 police-recorded road deaths for that year, matching the figure in MoRTH's annual series. WHO's modelled estimateModelled estimateA statistical estimate produced from multiple data inputs rather than a direct list of recorded deaths.The World Health Organization's 2021 figure has an uncertainty interval and cannot be transferred to 2024. for the same year is 216,618 deaths, with a 95 per cent uncertainty interval running from 193,271 to 239,965. The lower end of that interval still exceeds the police count.

The comparison points to substantial incompleteness in the police record. But the difference between the two figures is not a list of identified missing people: WHO's higher figure is a modelled estimate with uncertainty. It cannot be used as a multiplier on MoRTH's 2024 count. The ministry also notes that deaths may not flow back from hospitals into police records when the two systems are not linked. It assembled the national report from consolidated police returns, rather than directly from its electronic accident database. These limits affect how literally every chart on this page can be read.

What can these figures say about prevention?

The report is detailed enough to direct sharper questions. Why are so many of the victims people on two-wheelers or on foot? What is happening on the roads where rural deaths are recorded? Why do some police categories contain so many cases? The official tables establish scale and location, but they do not establish which intervention would have prevented a particular death. A recorded traffic violation alone does not prove what caused a death. Nor does helmet non-use prove that every death in that group was preventable.

Better measurement is part of the response. Linking crash, hospital and death records would clarify how many people die after the initial police report. Counting journeys or kilometres by road user and road type would allow risk to be compared on a common exposure base. Local records could join age, victim type, vehicle, road design and injury outcome instead of being inferred from separate national margins. Until then, the annual report is a detailed map of police-recorded harm with important blank spaces.

How to read these numbers: methodology and caveats

The main source is MoRTH's Road Accidents in India 2024, compiled from calendar-year returns of state and union-territory police. Every observation plotted here was checked against rows in the original PDF, and the published category totals were reconciled. Tables 1.1, 1.5, 1.6, 2.1, 3.3, 4.2, 4.3, 4.4, 5.6, 7.1 and 7.3 supply the chart data. The complete model-generated transcription of the 260-plus-page report remains a draft; the cells selected for this article have their own source audit.

Unless stated otherwise, percentages in the death charts divide each group by the 177,175 people recorded as killed in 2024. The severity line divides deaths by reported accidents in each year and multiplies by 100. The separate 33.7 per cent fatal-accident share divides 164,378 fatal crashes by all 487,707 reported crashes in 2024. The time-of-day chart divides accident counts by 487,707 reported accidents; it does not count deaths. The rural accident share divides crashes recorded in rural areas by all reported crashes. The 2.14 per cent national-highway road-length share divides national-highway kilometres by all road kilometres as of March 2022. Neither is a travel-exposure rate. The selected top-five state chart leaves out every other state and union territory. Published age bands are treated as separate categories even though they share endpoints in print, and unknown ages remain unknown.

The WHO comparison comes from its India country profile in the Global status report on road safety 2023 and refers only to 2021. MoRTH's Table 1.8 gives rates relative to population and registered vehicles, but the report does not pair its death counts with the number of journeys or kilometres travelled. None of these tables establishes a single cause for a crash or a death.