To test the safety and efficacy of tirilazad mesylate, a nonglucocorticoid 21-aminosteroid, in improving the outcome of patients with aneurysmal subarachnoid hemorrhage (SAH), 902 patients were enrolled in a prospective randomized, double-blind, vehicle-controlled trial at 54 North American neurosurgical centers. Five patients were excluded prior to receiving any study drug. Of 897 patients who received at least one dose of study medication, 300 received a placebo containing a citrate vehicle, 298 received 2 mg/kg per day tirilazad, and 299 received 6 mg/kg per day tirilazad, all administered intravenously beginning within 48 hours of the SAH and continuing through 10 days posthemorrhage. All patients were also treated with orally administered nimodipine. At 3 months post-SAH, there were no significant differences (p < 0.025) among the groups with regard to mortality rate, favorable outcome on the Glasgow Outcome Scale, or employment status. During the first 14 days after the SAH, there were no significant differences among the groups in the incidence or severity of clinically symptomatic or angiographically identifiable cerebral vasospasm. Mortality data stratified by gender and neurological grade on admission (assessed according to a modified World Federation of Neurological Surgeons scale) demonstrated that the men with Grades IV to V had a 33% mortality rate in the vehicle group, 52% in the 2 mg/kg per day tirilazad group (p = 0.29), and 5% in the 6 mg/kg per day tirilazad group (p = 0.03). Tirilazad was well tolerated at both dose levels. Tirilazad mesylate at dosage levels of up to 6 mg/kg per day for 8 to 10 days following SAH did not improve the overall outcome in patients with aneurysmal SAH in this trial. The differences in the efficacy of tirilazad in this trial and a previously reported trial in Europe, Australia, and New Zealand, in which dosage levels of tirilazad of 6 mg/kg per day reduced mortality rates and increased good recovery, may be a result of differences in admission characteristics of the patients and/or differences in management protocols, including the use of anticonvulsant medications.