In a retrospective study, we compared the incidence and risk of mortality associated with CMV disease in adult allogeneic BMT and PBSC recipients who received ganciclovir prophylaxis three-times-per-week (78 patients) vs five-times-per-week (137 patients). Active CMV infection occurred in 28 (41%) and 26 (21%) in the three- vs five-times-per-week groups, respectively (P < 0.005). CMV disease developed in 11 (16%) and five (4%) patients who received ganciclovir prophylaxis in the three-times-per-week vs five-times-per-week groups (P < 0.004). The CMV-attributable mortality rate was 1.5% and 12% in the five- vsthree-times-per-week groups, respectively (P < 0.003). Risk factors for CMV disease, significant at the P < 0.05 level in the multivariate analysis, were ganciclovir prophylaxis at three-times-per-week, receiving a T cell-depleted (TCD) marrow, and tacrolimus as prophylaxis for GVHD. These data suggest that ganciclovir five-times-per-week significantly reduced the incidence and mortality of CMV disease in allogeneic BMT and PBSC recipients. However, ganciclovir five-times-per-week was less effective for the prevention of CMV disease in patients receiving TCD marrow or tacrolimus.