Cytomegalovirus (CMV) surveillance and preemptive therapy is the most commonly used strategy for CMV disease prevention in hematopoietic cell transplantation recipients. In 2007, we introduced a CMV prevention strategy for patients at risk for CMV disease using quantitative PCR surveillance, with treatment thresholds determined by patient risk factors. Patients (N = 367) received preemptive therapy either at a plasma viral load of ≥500 copies/mL, at ≥100 copies/mL if receiving ≥1 mg/kg of prednisone or anti-T cell therapies, or if a ≥5-fold viral load increase from baseline was detected. Compared with patients before 2007 undergoing antigenemia-based surveillance (n = 690) with preemptive therapy initiated for any positive level, the risk-adapted PCR-based strategy resulted in similar use of antiviral agents, and similar risks of CMV disease, toxicity, and nonrelapse mortality in multivariable models. The cumulative incidence of CMV disease by day 100 was 5.2% in the PCR group compared with 5.8% in the antigenemia group (1 year: 9.1% PCR vs 9.6% antigenemia). Breakthrough CMV disease in the PCR group was predominantly in the gastrointestinal (GI) tract (15 of 19 cases; 79%). However, unlike CMV pneumonia, CMV GI disease was not associated with increased nonrelapse mortality (adjusted hazard ratio, 1.19; P = .70 [GI disease] vs 8.18; P < .001 [pneumonia]). Thus, the transition to a preemptive therapy strategy based on CMV viral load and host risk factors successfully prevented CMV disease without increasing the proportion of patients receiving preemptive therapy and attributable toxicity. Breakthrough disease in PCR-based preemptive therapy occurs at a low incidence and presents primarily as GI disease, which is more likely to be responsive to antiviral therapy.