Abstract: The serum level of S100B has been proposed as an indicator of prognosis in head trauma; however, there are scant reports in this regard. This study aimed to investigate the prognostic value of serum S100B in patients with severe head trauma. In a cross-sectional, prospective study, 72 patients with severe head trauma (Glasgow coma scale = 8) were recruited from Tabriz Imam Reza Teaching Center from September 2009 to December 2011. Serum level of S100B was measured on admission, 48 h later and at discharge/decease by enzyme-linked immunosorbent assay (ELISA) method. According to their in-hospital outcome, the patients were categorized as discharged (group A, n = 42), expired due to cerebral injury (group B, n = 16) and expired due to non-cerebral causes (group C, n = 14). The three groups were comparable in terms of age and sex. While the mean level of serum S100B was not significantly different between the three groups on admission and at the time of discharge/decease, it was significantly higher in group B (2.4±1 μU mL-1) than in groups A (1.0±0.5 μU mL-1) and C (1.5±0.5 μU mL-1) 48 h after discharge (p = 0.001). Serum S100B levels on all three occasions were significantly correlated with Glasgow Coma scale in a reverse fashion. According to the Marshall grading system, the brain trauma was of grade I, II, III, IV and V in 18, 16, 13, 13 and 12 cases, respectively. The mean level of the serum S100B was significantly higher in grade V only when it was measured 48 h after admission (p = 0.02). In conclusion, serum level of S100B is a significant predictor of in-hospital mortality in patients with severe head trauma, when it is measured after 48 h of admission. It is also correlated with radiologic grade of the brain injury in such cases.