BACKGROUND: This study was undertaken to determine the incidence of preventable trauma death at a non-designated trauma centre in Singapore. METHODS: A retrospective audit was carried out on all trauma deaths that occurred between January 1993 and December 1994 at the National University Hospital, Singapore. Of the 138 deaths, 38.4% (53/138) of patients were dead on arrival and were omitted from the study. Data from the remaining 85 deaths were summarized and presented before a multidisciplinary review board and injuries were scored according to the Abbreviated Injury Scale (AIS). RESULTS: Except for one patient, all deaths had an AIS score of 16 or greater. Sixty-one per cent (52/85) of deaths were the result of severe head injuries, and the rest had severe injuries in more than one body region. Following the guidelines of the Trauma Research and Education Foundation of San Diego, the incidence of 'not preventable', 'potentially preventable' and 'frankly preventable' deaths were 77.6, 15.3 and 7.1%, respectively. The most common errors in management were caused by delays in inter-departmental transfer (25.9%) and missed initial diagnosis (16.5%). After assessment by the Coroner's Office, autopsies were carried out in 60% of the cases and yielded information that altered the assessment of preventable deaths, particularly in the group with multiple injuries. CONCLUSIONS: The present study concludes that the preventable death rates in the National University Hospital are comparable to those in non-trauma designated centres elsewhere. There can only be a decrease in the preventable death rate if an improved system of pre-hospital trauma care, improved interdisciplinary communication, closer supervision of the initial attending physicians and stricter protocols on clinical monitoring are established.