The relationship between the past history of selected diseases and the risk of dying from hepatocellular carcinoma (HCC) was analyzed using 110,792 cohort members (46,465 males and 64,327 females) recruited between 1988 and 1990 by the JACC Study (the Japan Collaborative Cohort Study for Evaluation of Cancer Risk). Significantly elevated hazard ratios (HRs) were observed in both genders for the past history of kidney diseases, liver diseases, gallstones or cholecystitis, diabetes mellitus, and blood transfusion. Further, when analyzed by age group (those 40-59 years of age were "younger" and those 60-79 years of age were "older"), although the significant associations were generally maintained, the magnitude of the HRs for liver diseases and diabetes mellitus seemed to be considerably different between the younger and older age groups for male cohort members. When the analyses were limited to cohort members without the past history of liver diseases, the past histories which had significantly elevated HRs were hypertension (HR = 3.14, 95% confidence interval (CI): 1.25-7.89), diabetes mellitus (HR = 4.17, 95% CI: 1.22-14.25), and blood transfusion (HR = 7.69, 95% CI: 3.09-19.15) in the younger male age group and gallstone or cholecystitis (HR = 2.58, 95% CI: 1.11-5.98) in the older male age group. On the other hand, for females, the significantly elevated HRs were gastric or duodenal ulcer (HR = 4.33, 95% CI: 1.09-17.25) in the younger age group and diabetes mellitus (HR = 6.16, 95%CI: 2.25-16.90) and blood transfusion (HR = 3.86, 95%CI: 1.58-9.41) in the older age group. However, since the evidence from our univariate analyses might not be decisive, multivariate Cox proportional hazards models controlling for potential confounders and effect modifiers will be required to obtain more valid or unbiased hazard ratios.