OBJECTIVE: To examine the CE of adding minipool NAT to current blood screening (CS) of volunteer blood donations to reduce the risk of HBV, HCV and HIV infection in the United States. METHODS: We developed a decision analytic model of screening volunteer blood donations in the US based on recently published Markov models of HBV, HCV, and HIV infection to estimate discounted lifetime costs and quality-adjusted life year (QALY) gains. Infection risk (including prevalence and the window period between antigen and antibody detectability in the donated blood), and test sensitivities were derived from the literature. Age-specific 10-year survival of transfusion recipients was from Vamvakas (1994) and the age distribution from a private managed care database for transfusions in 1995. Secondary analyses considered alternative screening strategies. RESULTS: The model estimated NAT would annually prevent 37, 128 and 7 transfusion-acquired cases of HBV, HCV, and HIV respectively compared to CS alone (6.2 million transfusion recipients). HCV had the greatest impact on total QALYs and costs. Although the cost per case of HIV avoided was 3-4 times that for HBV or HCV, the overall impact of HIV on CE was small. Adding NAT to CS would add 86 life years, at an incremental cost per life year gained of $2.1M and an incremental cost per QALY gained of $1.2M. The CS+NAT-p24 strategy dominated CS+NAT, and had an incremental cost per QALY of $0.9M compared to CS. Results were most sensitive to disease incidence rates, screening test costs, estimates of window period closure, and the age distribution of transfusion recipients. CONCLUSIONS: The CE of adding NAT to current screening, although not within a range considered cost-effective for health care treatments, may be reasonable when considered in the context of other blood-related preventive interventions such as autologous blood donation, and the desire for a zero tolerance level for infections from blood transfusions.