A prospective study to investigate risk factors for CMV disease was conducted in 94 renal transplant recipients. CMV disease was defined as either unexplained fever for greater than 3 days with viremia or unexplained fever for greater than 3 days with isolation of CMV from the urine or throat wash and at least one of the following: leukopenia, elevated serum alanine aminotransferase, or biopsy-proved invasive tissue infection of the lung or gastrointestinal tract. Fifty-three patients received immunosuppressive regimens consisting of prednisone and cyclosporine, with or without azathioprine. The remaining 41 patients were treated with these agents plus OKT3 (21 received OKT3 to treat rejection, 20 received OKT3 prophylactically). Thirty-seven patients were at minimal risk of CMV disease (donor and recipient seronegative for CMV); 12 patients were at risk of primary disease (donor seropositive, recipient seronegative), and 45 were at risk of reactivation disease (recipient seropositive at the time of transplantation). The incidences of CMV disease in the 3 groups were 0%, 58%, and 36%, respectively. Although the incidence of CMV disease in patients at risk of primary disease was not influenced by the immunosuppressive regimen, immunosuppression had a profound effect on the occurrence of CMV disease in CMV-seropositive transplant recipients. The incidence of CMV disease in those receiving OKT3 was 59%; but only 21% in those who did not receive OKT3. OKT3 increased the risk of CMV disease five-fold (odds ratio 5.2 (95% confidence limits 1.4-17.5)). In the CMV-seropositive patient, OKT3 was also the most important predictor of CMV disease by multivariate analysis (P less than 0.002). A pilot study of preemptive therapy with ganciclovir (2.5 mg/kg daily during OKT3 therapy) in 17 patients decreased the incidence of CMV disease without appreciable toxicity.