OBJECTIVE: To assess the efficacy of high-dose oral acyclovir therapy compared with preemptive, short-course ganciclovir therapy (administered only if cytomegalovirus [CMV] shedding occurred) to prevent CMV disease in liver transplant recipients. DESIGN: A randomized controlled trial. SETTING: Liver transplant center at a university-affiliated Veterans Affairs Medical Center. PATIENTS: 47 consecutive patients having liver transplantation. INTERVENTION: Patients were stratified by their CMV antibody status and the CMV antibody status of the donor and were randomly assigned to one of two treatment groups. Surveillance cultures for CMV (buffy coat and urine) were done every 2 to 4 weeks for 24 weeks in all patients. One group received high-dose oral acyclovir (800 mg four times daily). The experimental group received no acyclovir, but if surveillance cultures were positive, ganciclovir (5 mg/kg intravenously twice daily) was administered for 7 days. MEASUREMENTS: Cytomegalovirus shedding and CMV disease were measured in the two groups. RESULTS: Cytomegalovirus shedding before the onset of CMV disease occurred in 25% (6 of 24) of patients in the acyclovir group compared with 22% (5 of 23) in the experimental group. Cytomegalovirus disease developed in 29% (7 of 24) of the acyclovir group and in 4% (1 of 23) of the experimental group (P < 0.05). No hematologic toxicity occurred with ganciclovir. CONCLUSION: Oral acyclovir is ineffective prophylaxis against CMV in liver transplant recipients. Preemptive, short-course ganciclovir therapy in patients with CMV shedding was well tolerated and provided effective prophylaxis against subsequent CMV disease; this protocol targets the patients at risk for CMV disease and minimizes toxicity and expense.