Most crashes are one impact. A rollover is a dozen.
A car that hits a wall stops once. The energy goes in over a fraction of a second, in one direction, and the crash is over almost before it starts. That is the crash the seat belt and the airbag were built for. Front to back. Horizontal. One big deceleration that the restraints were engineered to spread out and slow down.
A rollover does not work that way. When a vehicle goes over, it does not stop. It keeps turning. Every quarter turn is a new event. Roof to the ground, then side, then hood, then the other side, and around again if the speed is still there. The body inside is loaded and unloaded and loaded again, from a different direction each time, for as long as the vehicle keeps rolling. There is no single impact to point to. There is a series of them, and the count is not always small.
Inside the cabin the physics turns cruel in a specific way. In a frontal crash the belt catches the strong bones, the pelvis and the chest, and holds the occupant against a horizontal force. In a roll the force is not horizontal. It is up, and it is out. The body is thrown toward the roof and toward the window on the outside of the roll. A belt built to hold you back does less to hold you down. Slack that never mattered in a front crash becomes room to move, and in a roll a few inches of room is the difference between staying inside the survival space and reaching the edge of it.
That is where the two signatures of the rollover come from. The first is the roof. When the roof meets the ground with the weight of the whole vehicle behind it, it can push inward, into the space where a head already is. The neck does not do well with a load driven straight down its length. It is the same mechanism as a dive into shallow water, the crown of the head meeting something that will not give, the spine taking a load it was never shaped to take. The catastrophic neck injury in a rollover usually comes from that instant, not from the tumbling itself.
The second signature is the window. In a roll the openings become the danger. An arm, a shoulder, a head can reach a window that a belt in a front crash would never have let them near. Partial ejection, a limb or the head outside the vehicle while it is still turning, is a rollover injury and almost nothing else produces it. Full ejection, the body leaving the vehicle entirely, belongs to the unbelted, and the roll is the crash that finds them, because the sideways and upward forces are exactly the ones a lap and shoulder belt was least built to fight.
Then there is why some vehicles go over at all when others only slide. It is height, and it is load. A tall vehicle carries its weight high, and a loaded truck carries it higher still. The higher the center of the weight sits, the less it takes to tip. A hard steer, a drift onto a shoulder that grabs a tire, a curve taken with a full trailer, and the lateral force reaches the point where the vehicle stops sliding and starts turning over. The same physics that puts the truck on its side is the physics that starts the clock on everything above.
None of this is about who steered where. That is a separate question with its own rules. This is about the injury, and the injury in a rollover is not one blow. It is the roof coming down, the window coming open, and the body being asked to survive a crash that happens over and over before it finally stops.
On the numbers, see the National Highway Traffic Safety Administration’s Initiatives to Address the Mitigation of Vehicle Rollover, which reports that rollovers were about 8 percent of light vehicles in crashes but 31 percent of occupant deaths, that about 95 percent of single-vehicle rollovers are tripped by something like a curb, soft soil, or a guardrail, that 64 percent of those killed in rollovers were partly or fully ejected, and that the federal roof crush standard, FMVSS 216, requires a roof to bear a load tied to the vehicle’s own weight. On the roof and the neck, see Bidez and colleagues, Occupant Dynamics in Rollover Crashes: Influence of Roof Deformation and Seat Belt Performance on Probable Spinal Column Injury (Annals of Biomedical Engineering, 2007), which found in instrumented rollover tests that roof deformation preceded the peak neck loads and that belt loads fell sharply at that moment as the crushing roof pushed the occupant away from the restraint. This is general information, not medical or legal advice.
