OBJECTIVE: To determine whether the use of prophylactic recombinant human granulocyte colony-stimulating factor (filgrastim) reduces the frequency of nosocomial infections in patients with either acute traumatic brain injury or cerebral hemorrhage. DESIGN: Randomized, placebo-controlled, double-blind, multicenter phase II study. SETTING: Intensive care units of seven medical centers. PATIENTS: Patients with either acute traumatic brain injury or cerebral hemorrhage who were intubated within 6 hrs of admission and who were expected to be ventilated for >72 hrs. INTERVENTIONS: Patients were randomized to receive daily subcutaneous injections of placebo (n = 21) or one of two doses of filgrastim (75 microg [n = 20] or 300 microg [n = 20]) for 10 days or until the absolute neutrophil count was >75,000 cells/mm3 or until extubation. MEASUREMENTS AND MAIN RESULTS: End points included increase in absolute neutrophil count, safety of filgrastim, and frequency of nosocomial infections (pneumonia, bacteremia, and urinary tract infection). Filgrastim caused a dose-dependent increase in absolute neutrophil count. There were no differences in the frequency of pneumonia or urinary tract infection; however, there was a dose-dependent decrease in the frequency of bacteremias (p < .05). Adverse events were similar among the three groups. There was one case of acute respiratory distress syndrome in the placebo group. CONCLUSION: In this patient population, use of filgrastim was safe and the agent appeared to reduce the risk of primary bacteremias but had no beneficial effects on mortality, length of stay, or other nosocomial infections.