This is a retrospective study on the outcome using the TRISS methodology of 94 significantly injured patients over a 24-month period, managed by the Hospital Trauma Team in a general hospital since the formation of the Team in August 1994. There were 37 deaths and nine (24.3%) of these were 'potentially preventable' according to TRISS methodology. Seven of these nine 'potentially preventable or unexpected deaths' were transferred from a nearby district hospital where there was no acute operative facilities. There was no significant difference between the sex, age, mode of injury or Injury Severity Score between the direct admission and transfer-in cases and the M-statistic values of the two groups were similar. Five of the nine deaths happened in the first 4 months after the formation of the Trauma Team and the other four were scattered in the subsequent 20 months. The rate of preventable deaths was 50% (five out of 10 deaths) in the first 4 months, and was 15% (four out of 26) in the subsequent period. The probable causes for the 'potentially preventable trauma deaths' were delay owing to interhospital transfer, delay in activation of the trauma team, unidentified intra-peritoneal haemorrhage, failure to control haemorrhage and delayed or inadequate definitive operation. The evident improvement in the reduction of unexpected trauma deaths were likely associated with the success factors of the improvement of the multi-disciplinary cooperation including mutual understanding, simultaneous patient assessment, higher readiness to use diagnostic peritoneal lavage or ultrasonography to evaluate blunt abdominal trauma, earlier senior participation in patient care, shortening in response time of supportive facilities and a gradual cultural change towards dedicated trauma patient care. Further reduction in unexpected deaths can be expected if better prehospital triage by ambulance staff is attained to transfer trauma patients to the most appropriate instead of the nearest hospital.