The window period in hepatitis C virus (HCV) infection is still a major problem in ensuring blood safety. HCV RNA detection by nucleic acid amplification technology-based tests has contributed to reduce the infectivity of blood products, but it is expensive, time-consuming and affected by a high prevalence of false-positive results. The aim of this study was to assess the performance of a newly developed enzyme immunoassay for the detection of HCV core antigen and its suitability for use in the screening of blood units in order to identify infecting samples that do not contain specific antibodies. For evaluation of laboratory performance, different samples were selected: to evaluate specificity, we tested 2,586 sera from blood donors, 500 general population samples, and 58 "difficult sera". All samples were tested by two screening assays, and results were negative. To estimate clinical sensitivity, 103 HCV RNA-positive, anti-HCV-negative samples, 6 natural seroconversion panels, and 9 commercial seroconversion panels were tested. Intra- and interassay precision were determined on two HCV-RNA-positive, anti-HCV-negative sera. Seventeen (0.66%) blood donor samples, 2 (0.4%) general population samples, and 2 (3.44%) difficult sera were initially reactive; 3 sera were positive on repetition. These 21 samples tested by reverse transcription-PCR were negative. The clinical sensitivity calculated with seroconversion panels and seroconverted patient samples was very similar to PCR sensitivity: 95% of PCR-positive, antibody-negative samples contained detectable HCV antigen. Data on intra- and interassay precision showed dispersion indices with values of less than 10%. In conclusion, the HCV antigen assay showed high sensitivity and specificity and could become a useful means of improving the safety of blood and blood products.