Survival and mortality outcomes for trauma patients admitted to Liverpool Hospital, Sydney were analysed to determine the adequacy of trauma care. TRISS and ASCOT survival probabilities and peer review were utilised to determine if deaths were avoidable. Evaluation methods were compared for assessment of care. During the study period 2205 trauma patients were admitted, 518 of which fulfilled the study entry criteria. There were 38 deaths. The age and Injury Severity Score (ISS) of survivors was 34 +/- 18 years, 9.8 +/- 9 (mean +/- sd) compared to age and ISS for nonsurvivors 37 +/- 22 years and 45 +/- 22*, *p < 0.001. Peer review suggested that 32 deaths were non avoidable, 4 potentially avoidable and 2 were probably avoidable. TRISS and ASCOT survival probabilities were > 0.5 in 16 and 18 patients respectively. TRISS and ASCOT had low positive predictive value (25%) in identifying avoidable deaths. The Z Score was 1.79. The standardised mortality ratio (SMR) was 1.16. The Effectiveness (E) value for outcome was 0.91. Poor communication within the Area Trauma System was the greatest contributor to avoidable deaths. All trauma deaths need peer review rather than solely relying upon ASCOT and TRISS probabilities to identify "unexpected" deaths for detailed review.