The hypothesis that corticosteroid therapy is responsible for the striking improvement in survival of patients with systemic lupus erythematosus (SLE) was investigated with two approaches: Of 250 published papers on SLE, 52 were chosen for the first analysis because they included sufficient information on diagnostic criteria and survival but were not limited to patients selected for a particular target organ. From each article percent survival by series, average duration to death, and 1 and 5 year survival curves were abstracted. Each statistic showed linear improvement in survival since the 1930s without a significant change (P > 0.10) in slope for the time period following the introduction of corticosteroids. The second analysis examined the effect of corticosteroid therapy on 142 patients with SLE followed at the Massachusetts General Hospital between 1922 and 1966. Although the steroid use was conservative, the patients' survival, from year of entry, was comparable to the 52 abstracted series. A prognostic index was used to stratify patients admitted in the steroid era (after 1950) for life table analysis of survival with and without steroids. Steroids had no discernible effect on overall survival in low and medium risk groups. Use of steroids was associated with improved survival among high risk patients (P < 0.05).