A medical and engineering study was made of 1074 car accidents involving 2520 vehicle occupants. The injuries they sustained were correlated with details of the crash and with contact points inside the cars. Where possible the forces which were generated in the crash were estimated and related to the injuries. In all, 39 car occupants suffered femoral shaft fracture, with an incidence of 1.7% and 0.8% for front and rear seat occupants respectively. Of these, 31 had associated injuries to other regions of the body; these were the cause of 13 of the 14 deaths in this group. Associated injuries were more severe in car occupants who were not wearing seat belts but the incidence of femoral fracture was not significantly different. The mean velocity change (delta-V) causing femoral fracture was 26 mph (42 km/h) and there was a higher incidence of femoral shaft fracture when delta-V was over 30 mph. Estimates of forces needed to cause fracture were higher than those found in cadaver studies. The time to union of femoral fractures did not correlate with the severity of the crash but was longer (mean 19 weeks) than the average for other femoral fractures.