OBJECTIVES: Prehospital triage of trauma patients is of paramount importance because adequate trauma center referral improves survival. We developed a simple score that is easy to calculate in the prehospital phase. DESIGN: Multicenter prospective observational study. SETTING: Prehospital physician-staffed emergency system in university and nonuniversity hospitals. INTERVENTIONS: We evaluated 1360 trauma patients receiving care from a prehospital mobile intensive care unit in 22 centers in France during 2002. The association of prehospital variables with in-hospital death was tested using logistic regression, and a simple score (the Mechanism, Glasgow coma scale, Age, and Arterial Pressure [MGAP] score) was created and compared with the triage Revised Trauma Score, Revised Trauma Score, and Trauma Related Injury Severity Score. The model was validated in 1003 patients from 2003 through 2005. MEASUREMENTS AND MAIN RESULTS: Four independent variables were identified, and each was assigned a number of points proportional to its regression coefficient to provide the MGAP score: Glasgow Coma Scale (from 3-15 points), blunt trauma (4 points), systolic arterial blood pressure (>120 mm Hg: 5 points, 60 to 120 mm Hg: 3 points), and age <60 yrs (5 points). The area under the receiver operating characteristic curve of MGAP was not significantly different from that of the triage Revised Trauma Score or Revised Trauma Score, but when sensitivity was fixed >0.95 (undertriage of 0.05), the MGAP score was more specific and accurate than triage Revised Trauma Score and Revised Trauma Score, approaching those of Trauma Related Injury Severity Score. We defined three risk groups: low (23-29 points), intermediate (18-22 points), and high risk (<18 points). In the derivation cohort, the mortality was 2.8%, 15%, and 48%, respectively. Comparable characteristics of the MGAP score were observed in the validation cohort. CONCLUSION: The MGAP score can accurately predict in-hospital death in trauma patients.