We evaluated the utility of alanine aminotransferase (ALT) measurements in predicting the incidence of hepatocellular carcinoma (HCC) in a cohort of 667 adults with chronic hepatitis C virus (HCV) infection from a community-based population in Japan, between 1994 and 2003. Cox proportional hazards regression analysis was used to describe the relationship between prediagnostic levels of ALT and the rate of HCC, after adjusting for age and gender; hazard ratios (HRs) and 95% confidence intervals (CIs) were obtained. Over an average of 8 years of follow-up, 52 HCC cases were identified. A significant association between a 20 IU/L difference in higher ALT level and subsequent HCC incidence was observed (HR 5 1.2; 95% CI: 1.1, 1.3). An abnormal ALT level (35 IU/L) increased the HCC rate by 4-fold compared to a normal ALT level (HR 5 4.1; 95% CI: 2.1, 8.0). Among 551 subjects with at least 4 repeated measurements of ALT, those with persistently abnormal ALT levels (n 5 118) had a strong, significantly increased HCC rate compared to those with persistently normal ALT levels (n 5 296) (HR 5 23.2; 95% CI: 3.0, 178.5). This study demonstrates that elevated ALT levels, measured on an average of 8 years before HCC diagnosis, predict an increased rate of HCV-associated HCC in a community-based population and that utilizing serial measurements to identify persistent ALT abnormality may be useful in determining HCC risk. ' 2006 Wiley-Liss, Inc.