An analytical and laboratory evaluation of a newly-developed fully-automated third generation ELISA for the detection of anti-HCV (Cobas Core Anti-HCV EIA, Roche) was undertaken. Coefficients of variation (CVs) calculated on positive control and serum samples ranged from 5.9 to 9.8% in the intra-assay precision test and from 3.9 to 11.3% in the inter-assay evaluation. With regard to the study of clinical laboratory performance, five groups of sera pre-screened with two third generation ELISA (Ortho HCV 3.0 ELISA; Innotest HCV Ab III) were assayed: anti-HCV negative samples (n = 932); anti-HCV positive samples (n = 449); difficult sera of different origin (n = 113); sera with discrepant results in the two ELISAs (n = 50); sera with an indeterminate result in one or more confirmatory test (n = 34). The overall concordance between the Roche anti-HCV EIA and the two reference assays was 97.5 and 97.8% for the Ortho and for the Innogenetics assays, respectively. Although it is not possible to provide absolute figures for clinical sensitivity and specificity, the results of the study on discrepant samples show that the Cobas Core Anti-HCV gives a number of negative results with positive or indeterminate confirmatory anti-HCV tests, which is intermediate between the Ortho and the Innogenetics assay. In contrast, only 5% Cobas Core Anti-HCV reactive sera are not positive or clear-cut single band reactive by supplemental assays. The results show that the new fully-automated third generation anti-HCV test is a valid alternative to other commercially available assays for screening of antibodies to the hepatitis C virus.