Head-on crashes made up two per cent of US collisions in 2005 yet ten per cent of the fatal ones. Here is how collisions are sorted and how investigators work out what happened.
Road collisions fall into five common types: lane departure (including head-on), junction crashes (rear-end, angle and side impacts), pedestrians and cyclists, animals, and buildings. Reconstruction teams then use physics, skid marks, event data recorders, video and software to work out what happened.
In 2005, head-on crashes made up only two per cent of all US collisions, yet they accounted for ten per cent of the fatal ones. That gap is the central puzzle of crash typology: how a crash is classified says a great deal about how likely it is to kill, and about what investigators should look for afterwards.
Sorting collisions into head-on, side-impact, single-vehicle and multiple-vehicle types lets statisticians count each crash once and compare them. Reconstruction is the forensic craft that works backwards from skid marks, data recorders and video to establish what a driver actually did.
What are the main types of car collisions?
Road traffic collisions generally fall into five common types. Lane departure crashes happen when a driver leaves the lane and hits another vehicle or a roadside object, and they cover both head-on and roadway departure crashes. Junction collisions include rear-end crashes and angle or side impacts. The remaining three involve pedestrians and cyclists, animals, and buildings.
Rollovers are not common, but they bring a greater rate of severe injury and death. Some are secondary events that follow a roadway departure or a collision with another vehicle. When several vehicles are involved the term "serial crash" may be used, and for many vehicles "major incident" tends to be preferred to "pile up".
Counting practices differ by country. California sorts collisions into head-on, sideswipe, rear end, broadside, hit object, overturn, auto-pedestrian, other, and a final bucket for crashes with no type recorded. French official statistics use seven types: two vehicles frontal, rear or side; three or more vehicles in series; three or more in multiple collisions; other collision; and without collision. The point of such lists is that each crash falls into one type and only one, so that statistics can be compared.
Head-on crashes were two per cent of US collisions in 2005 but ten per cent of the fatal ones.
Why is a head-on collision so dangerous?
A head-on collision happens when two vehicles travelling in opposite directions meet front to front. The usual trigger is a vehicle straying into the path of an oncoming one, though the root cause is sometimes a steering overcorrection after drifting towards the road edge.
The arithmetic is brutal. In a head-on crash, vehicle speeds are additive, so two cars each doing 45 mph meet at an effective 90 mph. If the cars weigh about the same, each set of occupants feels forces much like those from hitting a stationary car at 90 mph. If one vehicle is heavier, the physics stops being symmetrical, because kinetic energy depends on both speed and weight.
The closing speed is often blamed for the high death toll, but it is a misleading measure on its own. Two vehicles at 50 mph are equivalent to one moving vehicle hitting a stationary one at 100 mph, yet replace the parked vehicle with a solid wall or a bridge abutment and the equivalent speed drops back to 50 mph, except where a lighter car meets a heavier one.
Where head-on crashes happen
The likelihood is greatest on roads with narrow lanes, sharp curves, no separation between opposing traffic and heavy volumes. Severity rises with speed, which makes busy single-carriageway roads outside towns the riskiest of all. Motorways, despite high speeds, rarely carry high head-on risk because of median treatments such as cable barriers, concrete step barriers, Jersey barriers, metal barriers and wide medians.
The figures show how serious the problem is. In Canada in 2017, head-on collisions involved 6,293 vehicles and 8,891 people, injuring 5,222 and killing 377. In France in 2017 and 2018, 2563 and 2556 head-on collisions outside built-up areas and motorways killed 536 and 545 people, about 16% of all road fatalities. In Quebec, head-on collisions figure in 8% of work-related incidents, rising to 23% in rural zones with a limit above 70 km/h (43 mph).
Head-on, sideswipe and run-off-road crashes all belong to the lane-departure family, because they share causes and differ in consequences. A wrong-way entry crash, where a driver turns onto a motorway off-ramp instead of the on-ramp, is another variant, and designers put special signage at off-ramps to discourage it.
How are head-on collisions prevented?
The biggest gain comes from keeping oncoming traffic apart. Median separation can reduce road collisions by something in the order of 70%, and both Ireland and Sweden have run large safety-fencing programmes on 2+1 roads.
Barriers come in three kinds. Rigid concrete systems are the most common today and the costliest to install, but their life-cycle costs are low. Semi-rigid guardrail can cost as much as $100,000 per mile to install and is meant to absorb impact, so each crash and repair adds to its life-cycle cost. Flexible cable barriers are the most forgiving and cheapest to put up, but repairs after crashes make them expensive over time, even though calculated benefits reach as much as $420,000 per mile annually.
Sealing the roadside shoulder reduces overcorrection crashes. Where that is not possible, a "safety edge" can help. The paving machine gets an attachment that forms a bevelled edge at 30 to 35 degrees instead of a near-vertical one, which reduces the steering angle needed for a tyre to climb back onto the road. With a vertical edge, the angle required is sharp enough to cause a loss of control, and the car may veer into oncoming traffic or off the far side.
The European Road Assessment Programme's Road Protection Score measures lane separation as one of its main crash-type elements. Motorways typically earn a 4-star rating despite the highest permitted speeds. Star Rating research in Britain, Germany, the Netherlands and Sweden has found a pressing need for better median, run-off and junction protection on single carriageways at reasonable cost.
What is a side-impact or T-bone collision?
A side collision, known as a T-bone accident, means one vehicle takes a hit to its flank. Typical settings are junctions, car parks and multi-lane roads where vehicles pass each other. A related type, the sideswipe, is where the sides of two vehicles touch while travelling in the same or opposite directions. It usually causes less harm because the blow is glancing, although loss of control or a secondary crash can make it far worse.
A 2016 study found that in the EU, side impacts were much less common than frontal ones, at 22-29% against 61-69%. They tend to be far more dangerous for the people inside. Another EU report from 2015 attributed 35-40% of passenger fatality and serious injury to side impacts, against 55% to head-on crashes. A likely reason is how little car lies between the occupant and the striking vehicle. Even in the safest cars, fatal or serious injuries in side impacts typically occurred at 50 km/h (~31 mph), whereas frontal impacts needed 70 km/h (~43 mph).
Head-on and side impacts compared
| Measure | Frontal or head-on | Side impact |
|---|---|---|
| Share of EU collisions, 2016 study | 61-69% | 22-29% |
| Share of passenger fatal and serious injury, 2015 EU report | 55% | 35-40% |
| Typical speed for serious injury or death | 70 km/h (~43 mph) | 50 km/h (~31 mph) |
| Most serious injuries | Lower extremities (front and rear impacts) | Head and chest |
Side-impact figures and testing
In the United States in 2008, 5,265 people, or 22% of the 23,888 killed, died in vehicles struck in the side. About 9,700 died in side impacts in 2004. In Shanghai, 23% of 1097 serious accidents between June 2005 and March 2013 were side impacts, the leading collision mode there, and the head and neck were involved in about 64% of casualties.
Euro NCAP, IIHS and NHTSA test side impacts in different ways. As of 2015 all three run vehicle-to-vehicle tests, where heavier vehicles have lower fatality rates than lighter ones. NHTSA and Euro NCAP also run the harsher pole tests, where small vehicles fare no worse than large ones in fatality rate. Research indicates that the underbody is the best place to reinforce a car against intrusion by the pole. Side airbags became mandatory in the US in 2009, saving an estimated 1,000 lives per year.
Single-vehicle, junction, pedestrian and building crashes
A single-vehicle collision involves no other vehicle. The causes mirror head-on crashes, minus the oncoming car: roadway departure, fallen rocks or debris, rollovers and animals. Excessive speed, driver fatigue and drink or drugs are common factors, as are bad weather, poor drainage, narrow lanes and shoulders, insufficient banking and sharp curves.
Junctions produce a large share of crashes: about 50% of urban crashes and 30% of rural ones. Rural junction crashes are rarer but worse. According to NHTSA's Fatality Analysis Reporting System, 71% of fatal intersection crashes were in non-city areas, largely because of speed. Roundabouts are an effective, if expensive, way to slow traffic and cut high-speed right-angle collisions, while clear markings and signing are cheap ways to help.
Pedestrian deaths vary sharply by place. In the EU, pedestrians make up 22% of road deaths, from 11% in the Netherlands to 39% in Latvia and Romania, and 69% of them die in urban areas. Cyclists make up 8%, from 2% in Greece to 24% in the Netherlands. In Japan pedestrians account for 30% of road deaths. In France in 2014, 499 pedestrians were killed and 4,323 injured, 47% of the deaths coming at night. In the United States, pedestrian deaths rose 35% from 2008 to 2017, though areas with Vision Zero initiatives tended to buck the trend. The GHSA reports that 74% of US pedestrian deaths occur at night and 72% involve people not crossing at a provided crossing.
The injury sequence in a frontal pedestrian impact is well described: the bumper strikes the leg or knee, the lower body accelerates forward while the upper body rotates towards the car, damaging the pelvis and thorax, and then the head hits the windscreen. The US does not test vehicles for pedestrian safety.
Vehicle-into-building crashes are more frequent than most drivers assume. The Storefront Safety Council estimates about sixty per day in the US, leading to about 500 deaths per year, and a Texas A&M study put gas and convenience stores at about twenty per day.
How are traffic collisions reconstructed?
Traffic collision reconstruction is the process of investigating, analysing and drawing conclusions about the causes and events of a crash. Reconstructionists look at the driver, the vehicle, the road and the wider environment, using physics and engineering principles, sometimes with software for calculations and simulations. The work is done in cases involving fatalities or personal injury, often becomes the basis of expert testimony at trial, and feeds back into safer roads and vehicle design. Forensic engineers, specialised police units and private consultants carry it out.
The craft has a long pedigree. Crash analysis dates back almost to the first car crash, became more analytical in the 1930s, and in 1940 produced the first judicial opinion accepting speed calculated from skid length with the principle of conservation of energy. In 1985 NHTSA funded the first national training guidelines, which led to the Accreditation Commission for Traffic Accident Reconstruction, known as ACTAR. Hugh H. Hurt Jr. pioneered motorcycle crash research, showing that proper helmets reduced head injuries and that many crashes involved cars turning left in front of the rider.
Working the scene
Investigators visit the scene and examine every vehicle involved. They gather photographs, video, measurements, eyewitness testimony and depositions, and record how the car was steered, braked and lit, plus engine speed, cruise control and anti-lock braking. Tyre marks are physical evidence, and the length of a skid can often allow the original speed to be calculated. Because drivers frequently underestimate their own speed, an independent estimate is often essential. The road surface matters too, especially where traction was lost to black ice, diesel contamination or debris.
Documentation increasingly relies on 3-dimensional laser scanning. The scan produces a 3D point cloud that can be measured and turned into computer models, which feed straight into simulation programs and also help illustrate the analysis for witnesses and drivers.
The data inside the car
Modern cars carry their own witnesses. An event data recorder, often called a black box, saves data for a few seconds before, during and after a crash is detected. A commercial tool such as the Bosch CDR-Tool can download it, and the report typically lists speed, whether the brake was on, throttle position, belt use and steering angle. Some manufacturers, such as Tesla, provide their own retrieval tools.
Real-time telematics add another layer. A telematic control unit or aftermarket GPS tracker logs vehicle activity continuously. Standard systems may log every second, but some advanced devices record accelerometer readings at intervals as short as milliseconds, giving a "crash trace" that allows a slow-motion reconstruction. Video completes the picture: dashcams on consumer and commercial vehicles, and AI-driven video telematics in fleets that upload clips of harsh events to the cloud automatically.
From analysis to courtroom
Analysis means processing the data, testing hypotheses, building models and running simulations. Powerful, inexpensive computers have revolutionised the field, and the software ranges across computer aided drawing programs, vehicle specification databases, momentum and energy analysis, collision simulators and photogrammetry. Findings are compiled into reports, diagrams and animations so that juries can follow them. Animations deserve scrutiny, because software can display motions that are physically impossible; a reliable one rests on physical evidence and calculations that obey the laws of physics.
How are motorcycle crashes reconstructed?
The method uses the same principles of conservation of energy and momentum, plus knowledge of motorcycle dynamics and rider control. Investigators work in reverse, from the point of rest back to a moment before the crash when action could have prevented it. Five phases structure the work. Perception-reaction, estimated at 1.1 to 1.5 seconds, is when the rider spots a hazard and chooses a response. Avoidance is steering or braking with the front brake, rear brake or both. Pre-impact sliding follows, where locking the front wheel almost certainly means losing control. Then comes impact, then post-impact motion, as the rider often separates from the machine and travels on alone. Distances in both sliding phases help determine speeds.
Who investigates crashes?
The Royal Canadian Mounted Police run full-time forensic collision reconstructionists and analysts. In British Columbia they are known as ICARS, with units in each RCMP district. The force teaches Level-2 on-scene investigation, Level-3 advanced analysis and Level-4 forensic reconstruction at the Pacific Region Training Center in Chilliwack, and other police agencies can attend. The California Highway Patrol deploys Multidisciplinary Accident Investigation Teams, or MAITs, each made up of one sergeant as leader, two or more officers, one Motor Carrier Specialist I and one Senior Transportation Engineer from Caltrans.
What this means for someone driving
The statistics point in a few consistent directions. Speed amplifies everything: crash severity rises with it on every type of road. Rural junctions and undivided rural roads are where the worst outcomes concentrate. Side impacts injure occupants at lower speeds because there is so little structure beside the seat. And after any serious crash, the car's own recorder, a dashcam or a fleet tracker may hold the clearest record of what happened, since drivers frequently underestimate their speed.
Where is crash investigation heading?
Reconstruction is shifting toward digital evidence. Investigators increasingly draw on real-time telematics, and dashcams and video telematics have made video widespread as a data source. Road design is moving on its own track: Ireland and Sweden have already carried out large barrier programmes on 2+1 roads, and areas with Vision Zero initiatives have tended to resist the rise in pedestrian deaths seen across the United States.
