Ninety-two adult renal allograft recipients, receiving baseline immunosuppression with CsA and prednisone, were assigned randomly to one of the following regimens. CsA was discontinued (D/C group) in 47 recipients who were then maintained on Aza and prednisone; or Aza was added to continued low-dose CsA and prednisone (triple drug [TD] group) in 45 patients. Entry into the study required an absence of rejection and a stable creatinine for at least four months prior to randomization. The mean month of randomization was 8.34 +/- 2.9 for the D/C group, and 7.2 +/- 3.2 for the TD group. Following randomization, a significantly greater rate of rejection (P less than .01) was observed in the D/C group (40%) than in the TD group (13%). With a mean follow-up of 30 months, 41/47 of D/C allografts (87.2%) and 39/45 TD allografts (86.6%) were functioning. Nevertheless, rejection had a persistent adverse effect on allograft function, in both the D/C and TD groups, up to 36 months following randomization. Parameters such as donor-type and rejection prior to randomization did not identify recipients at risk for rejection following randomization. Therefore, although the CsA withdrawal regimen might be ideal, the opportunity to select appropriate candidates remained elusive. In contrast, the safety of the TD regimen became apparent. Neither significant nephrotoxicity nor hypertension was observed, and the opportunity for less daily prednisone was evident. Despite its additional cost, the TD regimen utilizing indefinite low-dose CsA, is preferred.