In the present study, the RIBA HCV serotyping SIA was evaluated with a cohort of injecting drug users. Serotyping may be a rapid and cost-effective method of determining genotypes in cohort studies. In this study, hepatitis C virus (HCV) antibody-positive sera from a cohort of 331 chronically infected injecting drug users, of which 167 were coinfected with human immunodeficiency virus (HIV), were serotyped by the RIBA HCV Serotyping SIA. Among the 331 specimens, serotype-specific antibodies were detected in 250 (sensitivity, 75. 5%), in which serotype 1 was predominant (57.2%), followed by serotype 3 (26.8%). Among the 331 specimens, 164 were HIV negative, and serotype-specific antibodies were detected in 151 (sensitivity, 92.1%), in which serotype 1 was predominant (59.6%), followed by serotype 3 (33.8%). For a subset of 58 samples taken from 19 chronically infected HCV seroconverters with a mean follow-up of 5 years, serotypes were compared with genotypes, which were determined by a line probe assay (HCV LiPa) and by direct sequencing of the products obtained by nested PCR of the 5' untranslated region. Among the 58 samples with known genotypes, serotype-specific antibodies were detected in 38 (total sensitivity, 65.5%), with a specificity of 78.9%. Thirty of these serotypeable samples revealed a serotype that corresponded to the genotype in the 58 samples (total positive predictive value, 51.7%). Of the 58 samples, 23 were coinfected with HIV, and when these were excluded, the total sensitivity increased to 76.5%, with a total specificity of 80.8% and a total positive predictive value of 61.8%. The serotyping assay showed a high total sensitivity (96.3%) for samples positive by HCV RIBA, version 3.0, with four bands. We conclude that the sensitivity of the RIBA HCV serotyping SIA is limited by the immunocompetence of the HCV-infected host. In general, samples from HIV-negative individuals containing genotype 1a had higher sensitivity, specificity, and concordance in the serotyping assay compared with genotyping, whereas samples containing genotype 3a were found to be more cross-reactive and untypeable. Therefore, the prevalence of genotypes other than genotype 1 could be underestimated if they are determined by serotyping, and improvements in specificity are recommended.