OBJECTIVE To evaluate the efficacy and safety of ganciclovir for prevention of cytomegalovirus (CMV) infection and disease. METHODS A randomized, placebo-controlled, double-blind trial. METHODS University-affiliated bone marrow transplant center. METHODS Cytomegalovirus-seropositive allogeneic bone marrow transplant recipients. METHODS Random assignment to receive either a placebo or ganciclovir at a dose of 2.5 mg/kg body weight every 8 hours for 1 week before transplant and then at a dose of 6 mg/kg once per day, Monday through Friday, after transplant when the post-transplant neutrophil count reached 1.0 x 10(9)/L. METHODS Cytomegalovirus infection (positive culture, seroconversion, positive histologic findings), CMV disease (pneumonia, gastroenteritis, the wasting syndrome), and study-drug toxicity. RESULTS Cytomegalovirus infection developed in 25 of 45 placebo patients (56%) but in only 8 of 40 ganciclovir patients (20%) (P < 0.001). Cytomegalovirus disease may also have occurred less often in the ganciclovir patients (4 of 40 patients [10%] versus 11 of 45 patients [24%]; P = 0.09). The probability of CMV disease occurring within the first 120 days after transplantation was 0.29 among the placebo patients but only 0.12 among ganciclovir patients (P = 0.06). Reversible neutropenia was the only appreciable toxicity related to ganciclovir and required interruption of the study drug after transplant in 25 of 43 ganciclovir patients (58%) and in 13 of 47 placebo patients (28%) (P = 0.005). Overall survival was similar in both the placebo patients (29 of 45 [64%]) and ganciclovir patients (28 of 40 [70%]; P > 0.2). CONCLUSIONS Prophylactic ganciclovir, started before transplant and continued after recovery of the post-transplant neutrophil count, reduces the incidence and severity of CMV infection in CMV-sero-positive bone marrow transplant recipients but is frequently associated with neutropenia.