Between September 26, 1980 and December 31, 1983, 230 splenectomized, transfused renal allograft recipients were randomized to treatment with either cyclosporine-prednisone (N = 121, 68 diabetic and 53 nondiabetic recipients; 73 cadaver and 48 related donor grafts) or azathioprine-prednisone-antilymphocyte globulin (N = 109, 61 diabetic and 48 nondiabetic recipients; 69 cadaver and 40 related donor grafts). The results were analyzed on March 31, 1984. Actuarial patient survival rates at 2 years were 88% in the cyclosporine and 91% in the azathioprine groups (p = 0.649). Graft survival rates at 2 years were 82% in all cyclosporine and 77% in all azathioprine-treated recipients (p = 0.150); the corresponding figures in the recipients of related donor grafts were 87% vs. 83% (p = 0.656), and in the recipients of cadaver donor grafts were 78% vs. 73% (p = 0.178). The 2-year graft survival rates were 81% in cyclosporine and 74% in azathioprine-treated diabetic recipients (p = 0.150) and 83% in cyclosporine and 81% in azathioprine-treated nondiabetic recipients (p = 0.604). Within the cyclosporine and azathioprine treatment groups, the differences in graft survival rates between diabetic and nondiabetic recipients were not significant (p = 0.822 and 0.423, respectively). Although there were no significant differences in graft survival rates, the cumulative incidence of rejection episodes within the first post-transplant year was significantly lower in the cyclosporine (34%) than in the azathioprine (60%) treated recipients (p = 0.001). In recipients of technically successful cadaver kidney grafts, the incidence of acute tubular necrosis (ATN) was 31% in cyclosporine and 30% in azathioprine-treated recipients (p = 0.822). Graft survival rates in azathioprine- and cyclosporine-treated recipients who did or did not undergo ATN were 72% vs. 89% (p = 0.011). The mean (±S.D.) serum creatinine levels (mg/dl) at 1 year were higher in cyclosporine (2.0 ± 0.6) than in azathioprine (1.5 ± 0.5) treated recipients (p = <0.001). A reduction in cyclosporine dose because of nephrotoxicity was required in 96 of the cyclosporine-treated patients (70%), and 25 were switched to treatment with azathioprine (21%). The incidence of all infections in cyclosporine-treated patients was approximately half of that in azathioprine-treated patients, and only nine per cent of the cyclosporine-treated patients were diagnosed to have cytomegalovirus infections during the first post-transplant year vs. 28% in azathioprine-treated patients (p = 0.002). Treatment of hypertension was required in 89 (74%) of the cyclosporine vs. 45 (41%) of the azathioprine-treated patients (p < 0.001), and hyperkalemia (55% vs. 16%) and hyperuracemia (52% vs. 11%) were also more common in the cyclosporine group (p < 0.001). The mean numbers of days the recipients were in the hospital at the time of transplantation were 15.6 ± 9.1 in the cyclosporine and 19.8 ± 10.7 in the azathioprine-treated patients, and the corresponding rehospitalization times were 21.3 ± 10.0 and 30.3 ± 15.5 days (p < 0.001). The results of our randomized prospective trial differ from those reported by other groups in which the graft survival rates were significantly higher in cyclosporine-treated recipients. The difference is primarily explained by the fact that azathioprine-treated recipients had a higher graft survival rate in our trial than in the other trials. Even though graft survival rates are not increased by the use of cyclosporine at our institution, we believe that the advantages of cyclosporine (fewer rejection episodes, fewer infections, and shorter hospitalization times) outweigh the disadvantages (higher serum creatinine levels and a higher incidence of hypertension), and that cyclosporine should be used for immunosuppression in most renal allograft recipients.