Forty-one patients with systemic lupus erythematosus and glomerulonephritis were studied in a randomized drug trial. Thirteen patients received prednisone only (Group 1), 16 received oral cyclophosphamide and oral azathioprine (1 mg/kg body weight . d of each initially) (Group 2), and 12 were given boluses of intravenous cyclophosphamide (0.5 to 1.0 g/m2 body surface area every 3 months) (Group 3). The mean observation period was 42 months (range 1 to 6.5 years). Renal function deteriorated in four of 12 patients in Group 1 and three of 27 patients in Groups 2 and 3 (p = 0.114). By life-table analysis, 86% of the entire group survived 5 years after entry to the study. Marked hypertension, fluctuating changes in serum creatinine, erratic changes in levels of antibody to DNA, reduced C3 levels, increasing proteinuria or sustained hematuria, and flares of extrarenal disease activity occurred more commonly in Group 1. Infectious complications were not commoner in Groups 2 and 3. We conclude that any marginal benefits produced by the programs tested cannot be shown in this class of patients without markedly increasing the sample size. Our current studies involve more vigorous treatment of patients with more acute disease and less treatment during prolonged periods of relatively good health.