The routine diagnosis of hepatitis C virus (HCV) infection is based on the detection of anti-HCV antibodies by two main methods (enzyme immunoassay [EIA] and chemiluminescence immunoassay [CIA]) but false-positives are a problem. We investigated three anti-HCV tests: two CIAs (Cobas e 601 and Architect i2000SR); and one EIA (Ortho HCV 3.0). Two other anti-HCV tests were also performed as supplementary and confirmatory tests, respectively: a recombinant strip immunoblot assay (RIBA HCV 3.0 SIA) and a reverse transcriptase polymerase chain reaction-based assay for HCV-RNA. After discriminating the false-positive results, the true anti-HCV seropositivity rate in 7156 serum samples was 0.91%. The seropositivity and false-positive rates for the Cobas e 601, Architect i2000SR and Ortho HCV 3.0 anti-HCV tests were 1.9% and 0.99%, 1.2% and 0.29%, and 0.87% and 0.01%, respectively. The mean level of HCV-RNA was 3399 x 10(3) IU/ml. Critical levels for false-positivity for HCV-RNA were a cut-off index of 200 for Cobas e 601, a signal/cut-off (S/CO) of 5 for Architect i2000SR and an S/CO of 1.2 for Ortho HCV 3.0. Positive and negative results for the RIBA HCV 3.0 SIA assay all accorded with the HCV-RNA assay, except for 23 (17%) 'indeterminate' results, all of which were negative with the HCV-RNA assay. In conclusion, to eliminate doubts related to false-positive findings in the initial HCV screening tests, additional confirmatory HCV-RNA assay should be performed.