BACKGROUND: The modified Rapid Emergency Medicine Score (mREMS) is a recently published index to estimate the severity of trauma patients; however, little is known about its performance in patients with different types of trauma. This study verified the predictive capacity of mREMS in-hospital mortality in patients of blunt and penetrating trauma with and without traumatic brain injury (TBI) and the performance of this index compared with the Rapid Emergency Medicine Score, Injury Severity Score, New Injury Severity Score, and Trauma and Injury Severity Score. METHODS: This is a retrospective, correlational study that analyzed trauma patients 18 years or older, who attended at a hospital in Rio de Janeiro, Brazil. The receiver operating characteristic (ROC) curve was applied in the analyses. RESULTS: The sample consisted of 987 patients, 359 (36.4%) with TBI (225 blunt and 134 penetrating trauma). Regarding mREMS, the area under the ROC curve for TBI patients for in-hospital mortality was 0.506 (95% confidence interval [CI], 0.404-0.609) for penetrating injuries and 0.486 (95% CI, 0.402-0.571) for blunt injuries; the values in patients without TBI were 0.629 (95% CI, 0.554-0.703) and 0.618 (95% CI, 0.552-0.684), respectively. In relation to the other indices the mREMS presented the lowest area under the curve/ROC for penetrating and blunt TBI, and the Rapid Emergency Medicine Score for extracranial injuries. CONCLUSION: The mREMS showed no prognostic capacity for patients with TBI, and it presented the worst performance in relation to the Injury Severity Score, New Injury Severity Score, and Trauma and Injury Severity Score to discriminate cases of in-hospital mortality when considering trauma patients with and without TBI.