BACKGROUND There have been no comparisons of the relative sensitivity of the two Food and Drug Administration-licensed multiplex (MPX) nucleic acid test (NAT) systems (Procleix Ultrio [Gen-Probe], TIGRIS platform [Novartis]; and cobas TaqScreen MPX [Roche Molecular Systems], cobas s 201 platform [Roche Instrument Center]) for detecting hepatitis B virus (HBV)-infected donors in minipool sizes (MP) used in the United States. STUDY DESIGN AND METHODS Routine blood samples from Thailand were obtained from plasma units from 129 hepatitis B surface antigen (HBsAg)-negative, HBV NAT-yield donations. Blinded US testing included antibody to hepatitis B core antigen (anti-HBc), NAT using both manufacturers' systems (undiluted-individual donation [ID], in singlet and diluted 1:6 and 1:16 in triplicate), quantitative antibody to hepatitis B surface antigen, HBV DNA viral loads, and HBV genotyping. HBV yields in the United States were estimated using the incidence/window period (WP) model and compared to the calculated assay sensitivities. RESULTS Eighty samples were classified as occult HBV (anti-HBc reactive) and 49 as WP (anti-HBc nonreactive). For US pool sizes, MPX detected significantly more samples than Ultrio (MPX MP6 vs. Ultrio MP16; p < 0.0001 for occult and WP). Ultrio MP16 results were not statistically different from Ultrio MP6 (p = 0.68 for occult; p = 0.42 for WP). There was no difference between platforms for MP sizes used in most of the world (MPX MP6 vs. Ultrio ID; p = 0.70 for occult and p = 0.34 for WP). Viral loads were higher in WP samples. Modeled yield estimates were consistent with measured assay sensitivity on the Thai donor samples. CONCLUSIONS As used in the United States, MPX MP6 is more sensitive than Ultrio MP16, but the impact of this difference is mitigated by low numbers of HBV WP infections.