OBJECTIVE Scores have been developed to assess the quality of trauma treatment. Our objective was to investigate prospectively whether there is a significant difference between the predicted survival and mortality rates with ISS, TRISS and ASCOT and observed rates in our patients. If the answer is affirmative, we also wished to evaluate whether the difference is significant enough to justify the greater data collection effort required for ASCOT as opposed to TRISS, and for TRISS as opposed to ISS. METHODS Charts from 470 of 484 blunt multiple trauma patients with an ISS of 8 or more documented as part of a National Fund Study between June 1990 and June 1996 were reviewed. We compared the survival and mortality rates calculated by ISS, TRISS and ASCOT with the observed survival and death rates. RESULTS Only the predicted survival and death rates calculated by ISS differed significantly from the observed rates. There were also significant differences between ISS and TRISS, and ISS and ASCOT, with regard to the predicted rates. CONCLUSIONS In our unit ISS, which is simple to use, is not suitable for quality assessment. TRISS and ASCOT are of equal value. We prefer TRISS until a better, internationally recognised score is available, since it involves less effort and is less error-prone. General recommendations for Switzerland cannot be made until further results from other national accident and emergency units are available.