Prophylactic antiviral treatment for prevention of CMV infection and disease could be valuable after UCBT, where the incidence of CMV infection is very high. The purpose of this study was to evaluate the efficacy and safety of a strategy of prophylaxis of CMV infection and disease in patients undergoing UCBT. From May 1997 to May 2005, 52 CMV-seropositive adults with hematologic malignancies (15 in advanced phases) underwent UCBT at our institution. CMV prophylaxis consisted of high-dose intravenous acyclovir from day −5 until engraftment followed until day +100 by intravenous ganciclovir (GAN, 5 mg/kg per day 3 to 7 days per week) in the first 38 patients and by oral valganciclovir (VALGAN, 900 mg per day as a single daily dose) in the last 14 patients. All patients received thiotepa, busulfan, cyclophosphamide and antithymocyte globulin as conditioning, cyclosporine and prednisone as GVHD prophylaxis, and filgrastim from day +7 until engraftment. All 52 patients were considered as evaluable for efficacy and those actually receiving the scheduled CMV prophylaxis with GAN or VALGAN were considered evaluable for toxicity. Median age was 33 yr (range, 18–47), HLA match was 6/6 in 2 (4%), 5/6 in 19 (37%), and 4/6 in 31 cases (60%), and median number of nucleated and CD34+ cells infused was 2.1 x 10E7/kg (range, 0.9–5) and 0.9 x 10E5/kg (range, 0.1–5.7) respectively. The cumulative incidence of CMV infection was 46% at day 100 and 54% at days 180 and 365, and the cumulative incidence of CMV disease was 2% at day 100, 8% at day 180 and 15% at day 365. No clear differences between patients taking GAN or VALGAN were observed in the cumulative incidence of CMV infection (45% and 50% at day 100 and 53% and 57% at days 180 and 365, respectively; P=0.89) and CMV disease (3% and 0% at day 100, 10% and 0% at day 180 and 16% and 14% at day 365, respectively; P=0.75). Twenty-two patients assigned to receive GAN and 8 patients assigned to receive VALGAN experienced 31 and 11 episodes of other severe infections respectively. GAN was withdrawn in 2 patients due to renal toxicity and VALGAN in 2 patients due to neutropenia. Ten patients (7 receiving GAN and 3 receiving VALGAN) experienced at least one episode of recurrent CMV infection. Two of the 8 patients who developed CMV disease, both with GAN, died from CMV. Higher doses of nucleated (P=0.008) and CD3+ (P=0.04) cells infused, presence of acute GVHD below grade II (P=0.01), and use of Thymoglobulin (P=0.008) were associated with a lower risk of CMV infection. The CD3+ cell dose was inversely associated with the risk of CMV disease (P=0.01). These results suggest that CMV prophylaxis with intravenous GAN or oral VALGAN is both safe and effective to reduce and/or delay the occurrence of CMV infection and disease after UCBT. Further, these data show, for the first time, the importance of the CD3+ cell dose infused and of other characteristics in the development of CMV infection and disease after UCBT.