Implementation of a regional trauma care system requires a field triage tool that identifies the severely injured patient and transports him to a trauma center, while preserving the flow of minimally injured patients to community hospitals. We prospectively tested the Trauma Score (TS) as a field triage tool and evaluated its accuracy against that of the Injury Severity Score (ISS), calculated after the patients' injuries were fully defined. During an 18-month period, 1106 patients admitted to the trauma center at San Francisco General Hospital had a TS determined in the field (TS1) and on arrival at the emergency department. A TS1 of 14 or less defined a subgroup of 222 patients in whom 93% of the deaths occurred. Using an ISS of 20 or more as an indicator of life-threatening injury, we determined the predictive value of TS1. There were 66 false-negatives (ISS, greater than or equal to 20; TS1, 15 or 16) and 107 false-positives (ISS, less than 20; TS1, less than or equal to 14). Using a prehospital TS of 14 or less as an indicator of serious injury, only 20% of a major urban trauma population would qualify for diversion to a trauma center.