Why Are India's Road Crash Numbers Not Going Down? Are We Diagnosing the Problem Wrong?
Why Are India's Road Crash Numbers Not Going Down? Are We Diagnosing the Problem Wrong?
India has spent two decades telling drivers to slow down, wear helmets, and fasten seat belts. Road deaths have risen 87% since 2005 regardless, and accidents have become steadily more lethal, not less. That does not prove these interventions are wrong. It raises a different, more uncomfortable question: are we investigating crashes deeply enough to actually know what caused them? The Ministry of Road Transport and Highways' own 2024 report offers a clue — accidents routinely classified as "human error" may in fact involve faulty road design, and the classification calling them human error is, in the Ministry's own words, only "prima facie" true.
| Report | Road Accidents in India – 2024 |
| Publisher | Ministry of Road Transport & Highways (MoRTH), Transport Research Wing |
| Key figures | 4,87,707 accidents · 1,77,175 fatalities · over-speeding cited in 70.74% of accidents |
| Source | morth.gov.in |
The Core Issue
Every year, MoRTH publishes a national picture of why Indians die on the roads. In 2024, that picture says over-speeding was cited in 70.74% of all accidents — up from 68.40% just a year earlier. Every other specific violation category — drunk driving, wrong-side driving, red-light jumping, mobile phone use — actually fell in the same period.
To be clear: speed is unquestionably a major risk factor, and no serious road-safety intervention can ignore it. The question this raises is not whether speeding kills — it clearly can. The question is whether recording "over-speeding" as the cause adequately explains why a particular crash occurred, or whether it has simply become the finding that requires the least investigation to allege and the least evidence to sustain. Those are very different claims, and India's data collection process, as MoRTH's own report describes it, is currently better equipped to support the first than to test the second.
Why This Pattern Matters
India's road accident data is not merely descriptive. It shapes national policy, state enforcement priorities, and infrastructure spending under programmes like the Safe System approach now guiding both domestic policy and international funding. If crash records are systematically better at capturing driver behaviour than at capturing road, vehicle, and system-level factors, every downstream intervention — speed limit laws, helmet enforcement, vehicle safety mandates — risks addressing only part of the picture, while contributing factors like a poorly designed junction or a defective vehicle part go undetected and unaddressed.
This is not a hypothetical risk. MoRTH's own long-term figures show road deaths rising from 94,968 in 2005 to 1,77,175 in 2024 — an 87% increase — while the number of accidents grew by only around 11% over the same period. Severity, measured as deaths per 100 accidents, climbed from 21.6 in 2005 to 36.3 in 2024.
This persistence of rising fatalities, despite two decades of speed-focused enforcement and the 2019 Motor Vehicles (Amendment) Act, does not by itself prove that speeding is being wrongly blamed — fatalities are shaped by many factors beyond enforcement, including vehicle kilometres travelled, motorisation, road exposure, and urbanisation. But it should prompt a genuine question: is behaviour-focused enforcement alone sufficient, or are important road, vehicle, and system-level causes being missed because they are never systematically investigated in the first place?
There is also a direct human cost. The same "cause" finding that goes into a police FIR or chargesheet often becomes central evidence in a Motor Accident Claims Tribunal (MACT) proceeding deciding who receives compensation, and how much. A finding shaped by what was easiest to record, rather than what was fully investigated, does not just distort statistics — it can shape whether a grieving family is fairly compensated.
What Even the Government's Own Report Admits
This is not an outside critique. MoRTH's 2024 report contains its own remarkably candid admissions:
- Data is "sourced primarily from the State/UT Police Departments" and is described as reliable "notwithstanding the availability of other sources such as hospitals and State/UT Transport Departments."
- "The data collected by the police is not always evidence-based, as police personnel attending the accident site may record details after returning from the spot, often relying on memory."
- "The data collected by the police may be influenced by the value judgements and personal biases of the reporting personnel."
- Table 10.2 of the report lists "limited photographic and forensic evidence" as a known challenge whose direct impact is "weak causality and liability assessment."
India has already built better technical architecture for this: the Electronic Detailed Accident Report (e-DAR), designed to bring together police, transport, highways, and health data through real-time, geo-tagged, evidence-based capture. e-DAR is being rolled out across states and is genuinely in use. Yet the same 2024 report states outright that "e-DAR data has not been used for compilation of this report," except in one limited case used only for validation. So the gap is not that India lacks the data infrastructure to investigate crashes properly — it is that the institutional architecture is not yet using that infrastructure to produce the headline statistics that shape national policy. The 70.74% over-speeding figure, and the rest of Table 3.1, still comes from the same conventional, memory-dependent police reporting process the Ministry itself critiques in the same document.
The Legal Machinery Behind the Numbers
Part of why this gap persists sits in criminal law. Road accidents in India are typically charged under a small set of Bharatiya Nyaya Sanhita (BNS) provisions — Section 281 (rash or negligent driving), Section 125 (causing hurt or grievous hurt), and Section 106(1) (causing death by a rash or negligent act) — each of which can be invoked on a threshold as thin as an investigating officer's assessment at the scene, without necessarily involving the kind of forensic reconstruction that would distinguish genuine recklessness from a road that was unsafe by design.
The result is a legal system that often begins — and frequently ends — with a police determination of rash or negligent driving, while the underlying crash may never be independently examined for road design, vehicle condition, or other contributing factors. That determination is real evidence. It is not, on its own, a scientific finding of cause.
Beyond a Single Cause: What Proper Crash Investigation Would Actually Examine
A crash rarely has just one cause. A rigorous Safe System approach — the same framework Bloomberg Philanthropies' Global Road Safety Partnership and India's own policy commitments increasingly reference — looks at contributing factors across four categories:
Driver
- Speed
- Impairment
- Distraction
- Fatigue
Road
- Geometry & sight distance
- Median design
- Signage & lighting
- Pedestrian crossings
- Road surface & work zones
Vehicle
- Brakes & tyres
- Steering
- Lighting
- Safety systems
System
- Enforcement
- Road maintenance
- Emergency response
- Design standards
- Licensing & training
India's crash records are not wrong to note driver behaviour. The problem is that driver behaviour is often recorded as the finding, without a comparable, systematic examination of the other three categories.
The question worth asking at a repeat-crash location is not only "why are drivers crashing here" — it is "what is this road doing, repeatedly, to the people who use it."
Your Rights / What You Can Do
- A "rash and negligent driving" finding in your FIR or chargesheet is evidence — not necessarily the final scientific word on what caused the crash. Courts have recognised that a criminal finding does not automatically control the outcome of a separate compensation proceeding.
- Compensation claims can be supplemented with independent evidence — accident reconstruction, road-condition documentation, vehicle inspection reports — where genuine doubt exists about the stated cause.
- Photograph the accident site, road conditions, and any visible road defects immediately (when safely possible) — this can matter later, given how thin the official forensic record often is.
- Ask, not assume, when a chargesheet attributes an accident to driver fault alone, particularly at known accident-prone junctions or stretches with a history of similar crashes.
PRAN's Perspective
PRAN believes India's road safety crisis cannot be solved by treating incomplete crash investigation as a data quality footnote. Every crash should answer two different questions — "who may be legally responsible" and "what actually caused this crash" — and India's system should stop assuming the first answer automatically supplies the second.
A standard checklist — road geometry, signage, visibility, lighting, road surface, vehicle condition, speed evidence, CCTV, photographs, and GPS/telematics data where available — should be mandatory for every fatal crash, not an optional add-on left to individual officer discretion.
Crash investigation should function the way aviation and rail-accident investigation already does in India: an independent process — police, road engineer, and where warranted, forensic expert — that establishes cause separately from the criminal proceeding that follows, not as a byproduct of it.
Every crash record should show, transparently, whether photographs, GPS location, road geometry, and vehicle inspection data were actually collected — and e-DAR's evidence-based data, not the conventional police format, should become the actual basis for MoRTH's annual national statistics.
Locations with repeated crashes should trigger a mandatory causation audit examining road design and engineering factors, not just renewed enforcement drives targeting driver behaviour at the same spot.
Where causation is genuinely disputed, tribunals should be able to call for accident reconstruction, road safety audit findings, vehicle inspection, CCTV, or e-DAR records — rather than treating the police chargesheet as the presumptive last word.
Conclusion
India does not lack the technology to investigate road accidents properly — e-DAR proves that. What it currently lacks is the institutional practice of using that technology to answer the causation question as rigorously as the liability question.
Until crash investigation is treated as seriously as crash prosecution, India's road safety numbers will keep reflecting how convenient a cause was to record, not necessarily how a person died.
Disclaimer: This article is intended for legal awareness and public policy discussion purposes only. It does not constitute legal advice.
For more legal-policy analysis and consumer rights advocacy, visit:
PRAN – Policy Research Action Network Foundation
www.publicrightaction.org
เคญाเคฐเคค เคฎें เคชिเคเคฒे เคฆो เคฆเคถเคों เคฎें เคธเค्เคค เคเคคि เคธीเคฎा, เคนेเคฒเคฎेเค เคเคฐ เคธीเค-เคฌेเคฒ्เค เคाเคจूเคจों เคे เคฌाเคตเคूเคฆ เคธเคก़เค เคฆुเคฐ्เคเคเคจा เคฎें เคฎृเคค्เคฏु เคฆเคฐ เคเคเคจे เคे เคฌเคाเคฏ เคฌเคข़ी เคนै। เคธเคฐเคाเคฐ เคी เค เคชเคจी 2024 เคฐिเคชोเคฐ्เค เคธ्เคตीเคाเคฐ เคเคฐเคคी เคนै เคि เคฆुเคฐ्เคเคเคจाเคं เคो "เคฎाเคจเคตीเคฏ เคญूเคฒ" เคฌเคคाเคจा เค เค्เคธเคฐ เคेเคตเคฒ เคช्เคฐเคฅเคฎ เคฆृเคท्เคเคฏा เคจिเคท्เคเคฐ्เคท เคนोเคคा เคนै, เคเคฌเคि เคธเคก़เค เคกिเค़ाเคเคจ เคฏा เคตाเคนเคจ เคฆोเคท เคैเคธी เค เคธเคฒी เคตเคเคนों เคी เคเคนเคฐाเค เคธे เคांเค เคจเคนीं เคนोเคคी। เคชुเคฒिเคธ เคฆ्เคตाเคฐा เคฆเคฐ्เค เคंเคเคก़े เคธ्เคฎृเคคि-เคเคงाเคฐिเคค เคเคฐ เคชूเคฐ्เคตाเค्เคฐเคน-เคช्เคฐเคญाเคตिเคค เคนो เคธเคเคคे เคนैं — เคฏเคน เคฌाเคค เคฐिเคชोเคฐ्เค เคธ्เคตเคฏं เคฎाเคจเคคी เคนै। เคญाเคฐเคค เคे เคชाเคธ e-DAR เคैเคธी เคฌेเคนเคคเคฐ, เคธाเค्เคท्เคฏ-เคเคงाเคฐिเคค เคคเคเคจीเค เคชเคนเคฒे เคธे เคฎौเคूเคฆ เคนै, เคฒेเคिเคจ เคฐाเคท्เค्เคฐीเคฏ เคंเคเคก़ों เคฎें เค เคฌ เคญी เคชुเคฐाเคจी เคชुเคฒिเคธ เคฐिเคชोเคฐ्เคिंเค เคช्เคฐเคฃाเคฒी เคा เคนी เคเคชเคฏोเค เคนो เคฐเคนा เคนै। เคฏเคนी เค เคงूเคฐी เคांเค เคฎुเคเคตเค़े เคे เคฎाเคฎเคฒों (MACT) เคฎें เคญी เคฌिเคจा เคुเคจौเคคी เคฆिเค เคธ्เคตीเคाเคฐ เคเคฐ เคฒी เคाเคคी เคนै। PRAN เคा เคฎाเคจเคจा เคนै เคि เคนเคฐ เคฆुเคฐ्เคเคเคจा เคฎें "เคाเคจूเคจी เค़िเคฎ्เคฎेเคฆाเคฐी เคिเคธเคी เคนै" เคเคฐ "เคตाเคธ्เคคเคตिเค เคाเคฐเคฃ เค्เคฏा เคฅा" — เคเคจ เคฆोเคจों เคธเคตाเคฒों เคे เค เคฒเค-เค เคฒเค, เคตैเค्เคाเคจिเค เคเคตाเคฌ เคนोเคจे เคाเคนिเค, เคคเคญी เคธเคก़เค เคธुเคฐเค्เคทा เคจीเคคि เคธเคนी เคฆिเคถा เคฎें เคเคे เคฌเคข़ เคธเคेเคी।