We assessed the correlation between hepatitis C virus replication and antibody responses toward hepatitis C virus core ((222-31, NS3 (C33C1, NS4 (5-1-1 and C100-3) and NS5 proteins in 59 virus carriers. The concentration of serum hepatitis C virus RNA was determined by a competitive reverse transcriptionpolymerase chain reaction assay. All 50 patients with high viremic levels of 2 lo6 copies/& had antibodies to C22-3 and C33C. Antibodies to 5-1-1, C100-3 and NS5 proteins were detected less frequently (p < 0.01) in 72% (36 of 50),78% (39 of 50) and 84% (32 of 38) of such patients, respectively. As for the nine patients with low viremic levels of < lo6 copies/&, antibodies to C22-3, C33C, 5-1-1 andNS5 proteins were detected in only one patient (ll%), which was significantly less than the frequency for highly viremic patients (p < 0.01). An- tibody to C100-3 was also found less frequently in only four patients (44%) (p < 0.05). Thus, only four (44%) of the nine low viremic patients tested positive for any antibody compared with all 60 highly viremic patients (p < 0.01). These results indicate that highly viremic carriers can be detected by the presence of hepatitis C virus antibodies, but a considerable proportion of low viremic carriers may not show any serological evidence of hepatitis C virus infection. (HEPATOLOGY 1994;19: 1360-1365.)