PurposeA secondary brain injury could appear after traumatic brain injury (TBI) due to neuroinflammation, oxidation and apoptosis. Higher levels of serum melatonin have been found on admission for TBI in non-surviving than in surviving patients. Thus, the objective of this study was to know serum melatonin levels during the first week of TBI in surviving and non-surviving patients, and to know if serum melatonin levels during the first week of TBI can be used to predict mortality. MethodsPatients with an isolated and severe TBI were included; that is, if they scored < 10 points in non-cranial aspects of Injury Severity Score and < 9 points in the Glasgow Coma Scale. We measured serum melatonin concentrations at days 1, 4 and 8 of TBI. Thirty-day mortality was the end-point study. ResultsLower serum melatonin levels were found in the surviving patients (n = 90) than in the non-survivors (n = 34) on days 1 (p < 0.001), 4 (p < 0.001), and 8 (p = 0.02) of TBI. Serum melatonin concentrations on days 1, 4, and 8 of TBI had an area under curve (95% Confidence Interval) for the prediction of 30-day mortality of 0.85 (0.77-0.91; p < 0.001), 0.82 (0.74–0.89; p < 0.001) and 0.71 (0.61–0.79; p = 0.06) respectively. ConclusionsThe new findings of this study were the presence of higher levels of serum melatonin on days 1, 4 and 8 of TBI in non-survivors than in survivors, and the ability to predict 30-day mortality for serum melatonin levels measured at these time points. However, more research is necessary to confirm our results.