Between 1985 and 1995, 9380 traffic accidents occurring in the area of Hannover, Germany, were analysed; 12428 individuals had been injured and 387 (3.1%) had sustained a pelvic-ring injury (AIS(PELVIS)>2). In 131 cases (34%), the injuries were further classified (Pennal and Tile) and a technical reconstruction made of the accident: 52% were type A, 27% type B and 21% type C injuries; 46% were in cars, 12% on motorised two-wheelers, 10% on bicycles and 1% in utility vehicles; 31% were pedestrians. Pelvic-ring injuries occurred in restrained vehicle occupants in accidents with a (delta)V of more than 30 km/h, whereas they occurred in a considerable proportion of unrestrained vehicle occupants, pedestrians and bicyclists at lower (delta)V or collision speed. The percentage of B- and C-type injuries increased in crashes with higher (delta)v or collision speed. In addition to further improvements of the passive safety, lower collision speeds or (delta)V would reduce or prevent pelvic-ring injuries. Due to the small number of occupants protected by airbags in this study, their protective effect for the pelvis could not be assessed. The reconstruction of pelvic-ring injury mechanism in traffic accidents is possible when technical and medical factors are considered.