All the cases and controls emerged from a cohort study held in the past years. They were male, more than 20 years of age at the time of their enrollment into the cohort study, with serum specimens taken and kept at -20 degrees C. Every death of primary liver cancer (PLC), occurred since then, as a 'case' (78 in total), was compared with no more than 4 matched controls chosen from cohort members who were still alive until the last follow-up. ELISA technique was used to test anti-HCV antibodies in the stored serum specimens of the cases and controls. The Results showed that anti-HCV prevalence rates were 33.3% (26/78) of the cases and 15.3% (40/262) of their matched controls, respectively (chi 2 = 11.86, P < 0.01, ORMH = 3.0, PAR% = 23.40%). The results suggest that HCV infection is another important factor in the aetiology of PLC in Guangxi, China. Because the serum specimens tested were drawn years before the occurrence of PLC deaths, our conclusion need not worry about the sequential problem, which occurs in most of other kinds of case-control studies.