OBJECTIVE: To study the incidence and nature of pelvic fractures in rapidly fatal automobile accidents. DESIGN: Retrospective. SETTING: County Medical Examiner's Office. PATIENTS: The files of 255 consecutive motor vehicle accident fatalities examined at the Jefferson County Coroner/Medical Examiner's office (study period 1996-1998) were reviewed. We correlated this information with our previous findings, derived from a review of 392 such cases (study period 1994-1996). RESULTS: Approximately 25% of decedents involved in rapidly fatal automobile accidents sustained pelvic fractures. In 93% of the cases, postmortem radiographs were available and suitable for scoring according to the Orthopaedic Trauma Association nomenclature. The distribution of pelvic fractures by type was type A, 16%; type B, 32%; and type C, 52%, with the most common pelvic fracture being type C1 (26%). Additionally, pedestrians and motorcyclists were twice as likely to sustain a pelvic fracture, and the severity of pelvic fracture type seemed to correlate with increasing speed of the automobile. No correlation between drug use or direction of impact and incidence or type of pelvic fracture was observed. Compared with published studies on survivors of automobile accidents, our data suggest that pelvic injuries may tend to be more severe in victims who do not survive to hospitalization. CONCLUSIONS: Our data indicate that current estimates about the mortality of pelvic fractures may be faulty due to exclusion of victims who fail to survive to hospitalization. This series suggests that an appreciation of the full spectrum of pelvic ring disruptions requires collaboration between orthopaedic surgeons and forensic pathologists.