We compared two commercial assays for quantification of serum hepatitis C virus (HCV) RNA to investigate whether pretreatment levels of serum HCV RNA could predict the outcome of interferon (IFN) therapy. The Amplicor HCV Monitor test is based on a single, combined reverse transcription and amplification reaction carried out by the Tth DNA polymerase using specific primers for the 5' untranslated (UTR) region. The Quantiplex HCV RNA 2.0 assay is based on specific hybridization of viral RNA by synthetic oligonucleotides complementary to the 5'-UTR and core regions of the genome, allowing equal quantification of the six major genotypes. Receiver-operating characteristic (ROC) analysis was employed to identify the best cut-off value (predicting patients who were non-responsive to treatment) with corresponding sensitivity and specificity values. Logistic regression analysis was performed using these cut-off values. We studied 133 consecutive patients with chronic hepatitis C enrolled in a prospecsustained responders was 5322 copies ml-1 by the Monitor test and less than 0.2 million equivalents ml-1 (MEq ml-1) by the Quantiplex assay; for the 115 non-responders/relapsers, the median viraemia was 83,125 copies ml-1 and 1.128 MEq ml-1 for the Monitor test and Quantiplex assay, respectively. Spearman's rank test gave a correlation of 0.63 between assays. The best predicting cut-off values were 22,134 copies ml-1 for the Monitor test and 0.330 MEq ml-1 for the Quantiplex assay; their respective sensitivities and specificities were 72% and 75% for Monitor and 67% and 83% for Quantiplex. By logistic regression analysis, the age and gender-adjusted odds ratios of high vs low HCV RNA levels, defining the risk of non-response, were 10.6 (CI 3.1-35.7) for Monitor and 14.3 (CI 4.3-47.3) for Quantiplex. The two assays had comparable sensitivity for serum HCV RNA but they identified different predictive cut-offs for non-response to therapy.