The aim of this study was to determine whether multicentre quality controls for the detectability of viral genomes could contribute to the improvement of diagnostic performance in the participating laboratories. The study was carried out during two successive rounds, during which 18 laboratories specialized in nucleic acid testing analyzed, through a polymerase chain reaction (PCR) assay, a common panel of GB virus C (GBV-C)/hepatitis G virus (HGV) RNA-positive and -negative samples. During the first round, the laboratories used either an 'in-house' PCR procedure or a partly standardized commercial test. After decoding the results of the first round, the procedures of the participating laboratories were compared in order to establish a consensus procedure deduced from those of the laboratories which provided the best results. During the second round, each participating laboratory could use the resulting consensus procedure, or its own procedure, or both. The results of this quality control study indicated that, whatever method used, even specialized and trained laboratories may give false-negative or false-positive results. The commercial assay did not guarantee a systematic high quality level of results. The striking heterogeneity of results observed among laboratories using the same commercial assay confirm that molecular biology methods need skilled technicians. The results of this quality control study suggest that full standardization of viral genome detection, including all steps of the procedure, is necessary and that the laboratories performing PCR should participate in repeated quality control studies, whatever technique is being used.