Safety and error reduction in medical care is crucial to the future of medicine. This study evaluates trauma patients dying at a level 1 trauma centre to determine the adequacy of care. All trauma deaths at a level 1 trauma centre between 1996 and 2003 were reviewed by an eight-member multidisciplinary death review panel. Errors in care were classified according to their location, nature, impact, outcome and whether the deaths were avoidable or non-avoidable. Avoidable deaths were categorized as potentially, probably and definitely avoidable. Between 1996 and 2003, there were 17 157 trauma admissions, including 307 trauma deaths. The mean patient age was 47.7 years +/- 24.8 years, mean injury severity score 38.1 +/- 19.6. Of all deaths, 69 (22.5%) were deemed avoidable. Of the avoidable deaths, 61 (88%) were potentially avoidable, 7 (10%) probably avoidable and 1 (1.4%) definitely avoidable. Avoidable deaths were associated with patients with increased age, lower injury severity score, admissions to intensive care unit, longer hospital stay and treatment by a non-trauma surgeon (P < 0.05). Of the 307 trauma deaths, 271 (89.3%) patients experienced a total of 1063 errors, an overall error rate of 3.5 per patient. The error rate in the non-avoidable group was 2.9 per patient and 5.3 per patient in the avoidable group (P < 0.0001). Most errors occurred in the resuscitation area. Age, severity of injury, hospital length of stay and care by a non-trauma surgeon are factors associated with avoidable deaths. A new approach to trauma and injury care is required.