BACKGROUND: Using prospectively collected data from Ulleval University Hospital in Norway, standard TRISS-based methods with case mix correction were compared with analysis based on ISS stratified data. METHODS: Reference data were The Major Trauma Outcome Study (MTOS) controlled sites, used for calculation of AIS 90 based TRISS coefficients. Present TRISS convention requires RTS scoring on hospital admission, excluding many severely injured patients intubated before arrival. Therefore, all Ulleval patients were RTS scored using prehospital data if needed. RESULTS: There was 6.6% of MTOS controlled sites patients (mortality rate 26.7%) that had been excluded before estimation of TRISS coefficients because of lack of data for Ps calculation. Analyses based on ISS stratified data included these patients and indicated significant better performance at Ulleval for blunt, but not for penetrating trauma. No TRISS-based analysis detected this difference. CONCLUSIONS: The RTS convention should be changed to reduce patient exclusion. Presently, stratified ISS based data should also be analyzed.