Between September 1980 and June 1984, 246 splenectomized, transfused renal allograft recipients were stratified according to presence of diabetes and donor source, and randomized to treatment with either cyclosporine (CsA)-prednisone (pred) or antilymphoblast-globulin (ALG--azathioprine (AZA)--prednisone. As of August 1986, mean follow-up is 47 months. Over all, actuarial patient survival is 84% and 83%, respectively at 4 years. Corresponding graft survival is 70% and 63% for CsA-treated and ALG-AZA-treated patients (NS). Within the subgroup of diabetic recipients of cadaver grafts, graft survival is 70% for CsA-treated and 53% for ALG-AZA-treated recipients (P = .035). In the CsA group, 71% required either a significant reduction in CsA dosage with the addition of azathioprine or a complete switch to azathioprine, mainly because of CsA-associated nephrotoxicity. Of those CsA patients switched at a mean time of 21.3 +/- 16.4 months posttransplant with mean serum creatinine of 2.40 +/- .67, current serum creatinine is 1.79 +/- .63. Current mean serum creatinine values are significantly greater for patients randomized to CsA-pred (1.73 +/- .60) vs. ALG-AZA-pred (1.49 +/- .59), P = .014, even though most CsA-treated patients were eventually switched. The causes of graft loss are not different between CsA and ALG-AZA randomized patients. In nondiabetics, rejection is the most common cause of graft loss (17/33), whereas in diabetics loss due to complications from overimmunosuppression or death from cardiovascular events is significantly more common (27/44) than corresponding losses in nondiabetics (6/33, P less than .05). Switching does not seem to influence the incidence or cause of graft loss. Since most patients started on CsA-prednisone are ultimately switched to triple drug therapy, the latter is now the preferred initial treatment modality.