Subheadline: An IIT-Delhi study suggests police data captures only a small share of bicycle crashes, while motorcycles account for most self-reported collisions.
Standfirst: A new IIT-Delhi study offers a rare view of bicycle crashes that do not enter official police records. Of 218 self-reported crashes recorded through a cyclist cohort, only 2.8% were reported to police. The findings also show that motorcycles were involved in half of the self-reported collisions, while injuries were concentrated in the lower limbs and hands. The evidence points to a road-safety problem shaped not only by dangerous incidents, but also by how cities record them. This analysis examines what the study reveals about Delhi’s cycling environment, why police data alone may be insufficient, and how the mismatch between official records and cyclist experience complicates infrastructure planning.
The most important finding in the IIT-Delhi study is not simply that bicycle crashes occur in Delhi. It is that most of them appear to remain outside the systems normally used to understand road safety. The study, titled “Characteristics and Incidence Rates of Self-Reported Bicycle Crashes in Delhi: A Cohort Study”, found that only 2.8% of bicycle crashes identified through its research were reported to police. That leaves a substantial gap between the incidents cyclists experience and the incidents visible in official records.
The research was conducted by Srishti Agarwal and Rahul Goel of IIT-Delhi’s Transportation Research and Injury Prevention Centre. It used on-road intercept surveys involving 844 cyclists at 27 locations in Delhi last year. Of those cyclists, 742 provided crash and exposure information in at least one follow-up and were included in a prospective cohort followed for an average of 4.8 months. The study therefore did not rely only on police files. It asked cyclists directly about crashes and their consequences.
That distinction changes the picture of road danger. Police-reported crashes are often treated as a practical indicator of where and how people are being injured. But reporting depends on several conditions: whether the incident is considered serious enough to involve the police, whether the people involved remain at the location, whether medical treatment is required, and whether a cyclist believes an official complaint will produce a useful outcome. The study’s comparison between self-reported and police-reported crashes shows that these processes do not produce the same account of cycling safety.
The two datasets differed particularly in the type of vehicle involved. Half of the 218 self-reported crashes involved motorcycles as the striking vehicle. Cars accounted for 25% and electric three-wheeled rickshaws for 11%. By contrast, police records contained a higher proportion of trucks and a lower proportion of motorcycles. The difference suggests that the crashes most visible to police may not be representative of the crashes most frequently experienced by cyclists.
The study links the prominence of motorcycles to the way road space is used in Indian cities. Motorcycles constitute a large share of urban traffic and, according to the researchers, often occupy kerbside lanes where cyclists usually ride, particularly during congestion. This creates a specific conflict at the edge of the carriageway, where cyclists and motorcycles may be competing for the same limited space. The finding is therefore not only about driver behaviour. It also concerns the absence or weakness of dedicated cycling space and the consequences of mixing different road users in a narrow roadside environment.
The research also found that only 2% of the crashes were single-bicycle crashes. These included loss-of-control events, such as sudden braking or manoeuvring to avoid a collision, as well as incidents involving mechanical failure, objects or potholes. Most of the recorded crashes therefore involved another road user. That pattern differs from findings cited in the study for Australia, Europe and the United States, where single-bicycle crashes account for 40% to 70% of bicycle crashes.
The researchers attributed at least part of this difference to lower cycling speeds in Delhi. The supplied findings do not establish that speed is the only explanation, nor do they provide a complete comparison of road design, cycling behaviour or reporting systems across the different settings. They do, however, show why safety interventions cannot simply be transferred from one urban context to another. The vehicles involved, the space available to cyclists and the way crashes are recorded all shape the safety problem.
The injury data adds another layer. All police-reported non-fatal crashes in the study resulted in injury and received medical treatment. Among self-reported crashes, 78% resulted in injury, and 67% of those injured sought medical treatment. Knee and lower-leg injuries represented 60.4% of injuries, followed by injuries to the wrist, hand and fingers at 27.8%. Head injuries accounted for 14.8%.
These figures indicate that the consequences of a crash cannot be assessed only through fatality counts or police cases. A cyclist may sustain an injury, seek treatment and still not generate a police record. The result is a form of administrative invisibility: incidents affect health and mobility without necessarily appearing in the data used by authorities to identify dangerous roads or prioritise interventions.
The study reported a crash injury rate of 28.6 per million kilometres cycled. It said this was 60 to 200 times higher than in Denmark, Germany and the Netherlands, countries described in the study as having extensive cycling infrastructure. The comparison is presented as evidence supporting the role of infrastructure improvements in reducing bicyclist injuries. At the same time, the study’s other findings show that infrastructure must be understood in relation to local traffic conditions. A facility that does not protect cyclists from motorcycles using kerbside lanes may not address the principal conflict identified in Delhi.
This is where the question of measurement becomes a planning question. If police data records a disproportionate share of truck-involved crashes while cyclists’ self-reports show motorcycles as the most common striking vehicle, planners working only from police records may assign attention to the wrong risk pattern. The issue is not that one dataset is automatically correct and the other is not. Police records capture incidents that enter the official system, while self-reported data captures a broader range of experiences, including crashes that may not trigger formal reporting.
A stronger road-safety evidence base would therefore need to distinguish between at least three categories: crashes reported to police, crashes resulting in injury or medical treatment, and crashes experienced by cyclists but not formally reported. The IIT-Delhi study does not provide a new administrative system for doing this, but it demonstrates why the categories should not be treated as interchangeable.
The institutional responsibilities are also divided. Police records are generated through law-enforcement processes. Medical treatment data sits within the health system. Road design and the allocation of kerbside space fall within the responsibilities of transport and civic authorities. Cycling research connects these sources by documenting what happens before an incident becomes an official case. Without such links, each institution sees only one part of the road-safety problem.
The study’s design offers useful detail but also sets boundaries around what can be concluded. It surveyed cyclists at 27 locations and followed a cohort for an average of 4.8 months. The findings therefore describe the crashes reported by participants in the study and compare them with police-reported cases. The supplied material does not establish that the results represent every cyclist, every road or every season in Delhi. Nor does it identify the specific intersections or corridors where the crashes occurred. Those limitations matter for translating the evidence into location-specific action.
Even with those limits, the numerical pattern is clear enough to challenge a narrow understanding of cycling safety. There were 218 self-reported crashes, but only 2.8% were reported to police. Motorcycles were involved in half of the self-reported cases. Nearly four in five self-reported crashes resulted in injury, and lower-limb injuries made up 60.4% of recorded injuries. These figures describe a system in which routine conflicts at the road edge can produce harm without producing a proportionate official record.
The larger urban question is whether Delhi’s transport system treats cycling as a mode with a measurable safety requirement or as an activity that can be accommodated within leftover road space. The study does not assess a particular cycle-track project or prescribe a single design solution. Its evidence instead identifies the conditions that any intervention would need to recognise: motorcycles operating near the kerb, frequent interaction with other vehicles, injuries concentrated in exposed parts of the body, and a large gap between crashes experienced and crashes reported.
The findings also complicate the meaning of “safe cycling” in a city. A lower number of police cases does not necessarily mean fewer crashes. It may partly mean that crashes are less likely to be reported. Similarly, the presence of medical treatment records does not by itself reveal where the road design failed or which vehicle movement caused the injury. Safety policy needs both incident measurement and an understanding of the environments in which those incidents occur.
What the IIT-Delhi study confirms is that Delhi’s bicycle-crash problem is undercounted when viewed only through police data. It also confirms that motorcycles are central to the self-reported collision pattern identified by the research, while injuries are common among those crashes. What remains uncertain from the supplied findings is how the pattern varies across neighbourhoods, road types, times of day and cycling purposes. Those are the questions that future research and official monitoring will need to address if Delhi is to measure cycling risk more accurately.
