Culture-based preemptive therapy with ganciclovir was shown to reduce the incidence of cytomegalovirus (CMV) disease after bone marrow transplantation (BMT). Culture techniques did not detect CMV in 12% to 13% of patients before the onset of CMV disease. In a prospective study, 71 patients either received preemptive therapy based on polymerase chain reaction (PCR) technique (37 patients) or on culture assays (34 patients). In both groups, therapy was continued until clinical signs disappeared and PCR negativity was documented. Twenty-two patients in the PCR group and 15 patients in the culture group received antiviral therapy. PCR allowed detection of the virus (median day, +32 v day +49; P = .006) and introduction of antiviral therapy (median day, +44 v day +54; P = .02) earlier than did culture assays. The incidences of CMV disease (2 of 37 v 8 of 34 in PCR group v culture group, respectively; P = .02) and CMV-associated mortality (0 of 37 v 5 of 34 in PCR group v culture group, respectively; P = .02) were decreased, and the duration of ganciclovir therapy (P < .001) was shorter in the PCR-monitored group. Incidence and median duration of severe neutropenia (less than 500/microL) were lower in the PCR group (two v eight episodes, P = .02; median duration, 1.5 v 5 days, P = .04), as was the incidence of nonviral infections during/after antiviral therapy (2 of 37 v 9 of 34; P = .012). Thus, preemptive therapy based on more sensitive detection methods such as the PCR assay reduces the incidence of CMV disease and CMV-related mortality. Additionally, stopping and withholding antiviral therapy in a PCR-negative patient is safe and allows reduction of the duration and side effects of antiviral therapy.