We conducted a multicenter randomized, double-blind, placebo-controlled trial of early short-term, high-dose methylprednisolone sodium succinate in 223 patients with clinical signs of systemic sepsis and a normal sensorium (112 received glucocorticoid and 111 placebo). Patients also received antibiotics and intravenous fluids. Glucocorticoid or placebo was administered intravenously by a bolus (30 mg per kilogram of body weight over 15 minutes) followed by infusion of 5 mg per kilogram per hour for nine hours. The average time between the diagnosis of sepsis and infusion was 2.8 hours. The principal end point was 14-day mortality, which was similar in the placebo (22 percent) and glucocorticoid (21 percent) groups (P = 0.97). Mortality was also not significantly different between those receiving placebo and those receiving glucocorticoid in subgroups with evidence of sepsis (21 vs. 19 percent), gram-negative bacteremia (27 vs. 7 percent), gram-positive bacteremia (18 vs. 26 percent), or all gram-negative infections (25 vs. 17 percent). Resolution of secondary infection within 14 days was significantly higher in patients receiving placebo (12 of 23) than in those receiving glucocorticoid (3 of 16) (P = 0.03), but mortality rates were similar in both treatment groups for those with unresolved infection (36 vs. 31 percent). We conclude that early high-dose glucocorticoid therapy does not reduce mortality significantly in patients with systemic sepsis who have a normal sensorium, and therefore should not be used as adjunctive therapy.