The presence of hepatitis C virus (HCV) negative strand RNA in extrahepatic compartments based on PCR detection assays has been suggested in many reports with a very heterologous detection rate (from 0 to 100%). In this study, we have analyzed the presence of HCV negative strand in hepatic (liver biopsies, n = 20) and extrahepatic (sera, n = 32; PBMC, n = 26 and fresh bone marrow cells, n = 8) compartments from infected patients with three different reverse transcriptase (RT)-PCR-based assays using primers located in the 5' noncoding region, with or without a tag selected to display different viral loads (10(5)-3 x 10(7) genomic equivalent/ml or gram) and viral genotypes (n = 5). Using synthetic as well as biological templates, we could document extensive artifactual detection of negative strand RNA, due to self priming and mispriming events, even either 5' noncoding region primer pair was used, whereas both artifacts were dramatically reduced (mispriming) or eliminated (selfpriming) using CAP-based RT-PCR assay. Mispriming artifacts were directly correlated to the titer of positive strand RNA present in the sample. Using the CAP-PCR assay, the presence of HCV negative strand RNA was found in 75% of livers (16:20) and only 8% of PBMC, independent of the genotype involved, but could not be documented in sera (0:32) and fresh bone marrow cells (0:6). These findings suggest that caution regarding the type of RT-PCR assay used and the level of HCV positive strand RNA present in the biological sample analyzed has to be taken to avoid false identification of viral reservoirs. The findings suggest that hematopoietic peripheral cells can support HCV replication, although in a very limited number of carriers.