We compared the incidence of de novo tumors developing in 1165 primary adult renal allograft recipients treated with azathioprine (AZA)-prednisone (Pred)-antilymphocyte globulin (ALG) (CONV group) with that in 722 patients receiving cyclosporine (CSA) as part of double (CSA-Pred), triple (CSA-Pred-AZA), or quadruple-therapy (CSA-Pred-AZA-ALG) protocols. Mean +/- SD follow-up was 9.5 +/- 6.4 years for the CONV group and 6.2 +/- 2.7 years for the CSA group. Overall, 124 patients (10.6%) in the CONV group and 34 patients (4.7%) in the CSA group developed malignancies, with nonmelanoma skin cancers and lymphomas comprising 55% and 13% of cancers in the CONV group and 65% and 3% of cancers in the CSA group, respectively. There were no significant differences in overall cancer (p = 0.41) or skin cancer (p = 0.97) incidence between non-CSA-treated and CSA-treated patients by Kaplan-Meier life-table analysis; however, CONV-treated patients demonstrated a higher incidence of lymphoma (p = 0.05). The mean (+/- SD) time to overall and skin cancer occurrence was significantly shorter in the CSA group: 37 +/- 22 versus 90 +/- 52 months (p < 0.001) and 40 +/- 24 versus 92 +/- 52 months, respectively. When the Cox Proportional Hazard Model was utilized to assess the relative importance of age, diabetic status, donor source, sex, and immunosuppressive regimen in determining cancer development, age > or = 50 years and nondiabetic status were significant independent prognostic indicators, while immunosuppressive regimen was not. Graft and patient survival were significantly greater in CONV-treated patients developing cancer than in those who did not.(ABSTRACT TRUNCATED AT 250 WORDS)