BACKGROUND Third-generation anti-hepatitis C virus (HCV) enzyme-linked immunosorbent assays (ELISA) are now implemented in most laboratories in Europe, but have not yet been fully implemented in the United States. METHODS Two ELISAs (Ortho 3.0 and Ortho 2.0, Ortho Diagnostics, Raritan, NJ) were compared by tests on various serum panels: A) blood donor samples (n = 530) that tested positive in first- or second-generation anti-HCV ELISA; B) samples from persons with chronic non-A, non-B hepatitis (n = 185); C) samples from multiply transfused patients (n = 79); D) samples from patients on hemodialysis (n = 473); and E) samples from Dutch random blood donors (n = 2153). RESULTS In panels A, B, and C, 247 (100%) of 247 polymerase chain reaction (PCR)-positive and 278 (100%) of 278 second-generation recombinant immunoblot assay (RIBA-2)-positive specimens were detected by Ortho 2.0 and 3.0 (sensitivity, 100%). In the sera of panel D, used to represent a group of patients with a high risk for HCV, no additional positives were found by Ortho 3.0. In panel E, of 2153 blood donor samples, 2 (0.1%) were positive in Ortho 2.0 and 8 (0.4%) in Ortho 3.0. Two samples that were positive in both Ortho 2.0 and 3.0 were also positive in RIBA-2; one was positive on PCR. From the 6 remaining Ortho 3.0-positive (Ortho 2.0-negative) samples, 1 was positive in RIBA-2 (isolated anti-c100) and 3 were positive in third-generation RIBA (1/3 isolated anti-c100, 2/3 isolated NS5). All 6 samples were PCR negative. In first-time donors, no difference in specificity was found. CONCLUSIONS The sensitivity and specificity of the Ortho 3.0 ELISA are comparable to those of the Ortho 2.0 ELISA.