Serum samples of 9691 male adults had been collected and frozen for a prospective study of hepatocellular carcinoma in Taiwan. After an average follow-up period of 4.6 years, testosterone levels in the stored serum were measured by radioimmunoassay using commercial kits for 35 cases of newly developed hepatocellular carcinoma, 63 hepatitis B surface antigennegative and 77 hepatitis B surface antigen-positive matched controls. Elevated testosterone levels were found to be associated with an increased risk of hepatocellular carcinoma. The association remained significant after the adjustment for effects of other hepatocellular carcinoma risk factors, including hepatitis B surface antigen carrier status, positivity of serum antibody to hepatitis C virus, cigarette smoking, alcohol drinking, past liver disease history, and dietary habits. The multivariate-adjusted relative risk of hepatocellular carcinoma for men with testosterone levels in the upper tertile was 4.1 (95% confidence interval = 1.3–13.2) compared with those having levels in the middle or lower tertiles (P = 0.016). The results consistent with those observed in animal experiments support the hypothesis that testosterone plays a role in the etiology of human hepatocellular carcinoma.