This study evaluated the performance of third-generation anti-HCV assays in blood donors who were positive by second-generation anti-HCV, and assessed any possible relationship between antibody patterns, HCV replication and liver damage. Fifty-two second-generation enzyme immunoassay-positive asymptomatic Italian blood donors were retested for anti-HCV by third-generation enzyme immunoassay and recombinant immunoblot assay (Ortho third-generation enzyme immunoassay, third-generation recombinant immunoblot assay), utilising recombinant C33c and NS5 and synthetic peptide C100 and C22 antigens, and for HCV-RNA by "nested" polymerase chain reaction with 5' region primers. Alanine aminotransferases were tested monthly for 6 months. Two out of 52 second-generation enzyme immunoassay-positive donors were third-generation enzyme immunoassay, third-generation recombinant immunoblot assay and HCV-RNA negative. Among 50 third-generation enzyme immunoassay-positive cases, two had a third-generation enzyme immunoassay optical density < or = 1: one was third-generation recombinant immunoblot assay and HCV-RNA negative, and the other was third-generation recombinant immunoblot assay "indeterminate" and HCV-RNA-positive. The remaining 48 cases had third-generation enzyme immunoassay optical density > 1: six were third-generation recombinant immunoblot assay negative (one HCV-RNA+ve), eight "indeterminate" (two HCV-RNA+ve) and 34 positive (22 HCV-RNA+ve). All "indeterminate" subjects reacted only to C22. HCV-RNA was positive in 22/34 cases with positive third-generation recombinant immunoblot assay (two or more Ags), 3/9 "indeterminate" and 1/11 negative. Alanine amino-transferases were abnormal in 13 cases with positive third-generation recombinant immunoblot assay, one was "indeterminate" and three were negative.(ABSTRACT TRUNCATED AT 250 WORDS)