A prospective (Part I) and a retrospective (Part II) study were used to determine the safety and efficacy of corticosteroids in the treatment of septic shock. In Part I, 172 consecutive patients in septic shock admitted over an 8-year period were treated with either steroid or saline: 43 received dexamethasone (DMP), 43 received methylprednisolone (MPS), and 86 received saline. The study was double-blind and randomized, and the three groups were compared for age, severity of shock, presence of underlying disease, and year of study. In the 86 saline-treated patients, the mortality rate was 38.4% (33/86); in the steroid-treated patients, it was 10.4% (9/86). With MPS the mortality rate was 11.6% (5/43), and with DMP it was 9.3% (4/43). Thus, overall mortality was significantly less in the steroid-treated group than in the control group. Further, there was no significant difference in mortality rate between the DMP- and the MPS-treated patients. In Part II, 328 patients were studied retrospectively. One-hundred sixty were treated without steroid, and 168 were treated with either DMP or MPS. Again, the two groups of patients were compared for severity of shock, underlying disease, age, and year of study. Mortality among patients treated without steroid was 42.5% (68/160) and among patients treated with steroid was 14% (24/168); there was no significant difference in mortality rate between DMP- and MPS-treated patients. In Parts I and II combined, complications occurred in 6% of steroid-treated patients with no significant difference between DMP- and MPS-treated groups.