A reverse transcription polymerase chain reaction (RT-PCR) assay was set up to amplify, from chronically infected patients, the recently discovered hepatitis C virus (HCV) 3'non-coding region (3'NCR). A panel of 149 samples was tested by RT-PCR for the 3'NCR. Two detection methods of amplified products were evaluated: ethidium bromide staining on 3% agarose gel electrophoresis and DNA enzyme immunoassay ("DEIA"). Results were compared with those obtained by amplification of the 5' non-coding region (5'NCR), i.e. the "Amplicor" HCV RNA qualitative assay. Genotype distribution of the 86 Amplicor-positive samples was subtype 1a: n = 15 (17.4%); subtype 1b: n = 32 (37.2%); subtype 2a/2c: n = 7 (8.1%); type 3: n = 25 (29%); type 4: n = 2 (2.3%); type 5: n = 1 (1.2%); not determined: n = 4 (2.3%). Sixty-three sera were HCV RNA-Amplicor-negative, 32 of which were from HCV-seronegative patients and 31 from HCV-seropositive patients. All seronegative samples were negative by both PCR methods. None of the Amplicor-negative samples from seropositive patients were positive by the 3'NCR assay. Forty-seven (54.7%) and 83 (96.5%) of the 86 Amplicor-HCV-RNA-positive samples were positive after ethidium bromide staining and by the 3'NCR assay using DEIA, respectively. The limit of detection by end-point dilution was lower with Amplicor. No difference between genotypes was detected for the 3'NCR RT-PCR, and a high degree of concordance was obtained between the Amplicor and the 3'NCR DEIA results (97.4%). Nevertheless, further studies are needed before the 3'NCR RT-PCR assay could be used instead of the 5'NCR RT-PCR for diagnostic purposes.