To determine whether corticosteroids are efficacious in severe septic shock, we conducted a prospective study of 59 patients randomly assigned to a methylprednisolone, dexamethasone, or control group. Patients were treated 17.5 +/- 5.4 hours (mean +/- S.E.M.) after the onset of shock, and 55 patients required vasopressor agents. Early in the hospital course, reversal of shock was more likely in patients who received corticosteroids than in those who did not. Four (19 per cent) of 21 methylprednisolone-treated, 7 (32 per cent) of 22 dexamethasone-treated, and none of 16 control patients had reversal of shock 24 hours after drug administration (corticosteroid groups vs. control group, P less than 0.05). Patients treated with corticosteroids within four hours after the onset of shock had a higher incidence of shock reversal (P less than 0.05). At 133 hours after drug administration, 17 (40 per cent) of 43 corticosteroid-treated patients had died, and 11 (69 per cent) of 16 control patients had died (P less than 0.05). However, these differences in reversal of shock and survival disappeared later in the course. Overall, 16 (76 per cent) of 21 patients receiving methylprednisolone, 17 (77 per cent) of 22 patients receiving dexamethasone, and 11 (69 per cent) of 16 controls in the hospital died. We conclude that corticosteroids do not improve the overall survival of patients with severe, late septic shock but may be helpful early in the course and in certain subgroups of patients.