To examine the relationship between health practice in combination with health examination data and mortality, a population-based prospective cohort study was conducted in 9 towns and villages of Gunma Prefecture, Japan. Public health nurses interviewed inhabitants about their health practices. This cohort consisting of 7,694 subjects aged 40 approximately 60 years old at baseline in 1993 was followed. During the ten-year follow-up period, 99 males and 80 females deceased. The relative risks (RRs) of some factors were estimated by the Cox proportional hazards model. RRs and 95% confidence interval with multivariate adjustment for all-cause mortality were observed significantly for age in years at the baseline (1.08, 1.04 approximately 1.12, p<0.01), GOT (1.02, 1.00 approximately 1.05, p<0.05), GPT (0.98, 0.96 approximately 1.00, p<0.05), and no proteinuria (0.36, 0.14 approximately 0.91, p<0.05) in male subjects. In females, RRs (95% confidence interval) of age in years at baseline, GOT, and gamma-GTP on mortality showed significance, which were 1.09 (1.05, 1.14, p<0.01), 1.03 (1.01, 1.05, p<0.01), and 1.01 (1.00, 1.01, p<0.05), respectively. When seven health practices and age in years at baseline were used as covariates, RRs (95% confidence interval) of age and no smoking on mortality showed 1.10 (1.06, 1.14, p<0.01) and 0.57 (0.37, 0.89, p<0.05), which were significant in males. In females, significant RRs (95% confidence interval) of age, no smoking and regular exercise on mortality were observed, which were 1.11 (1.07, 1.15, p<0.01), 0.48 (0.24, 0.94, p<0.05), and 1.63 (1.01, 2.63, p<0.05), respectively. These results suggested that increased risk of death was independently associated with aging, GOT, and smoking in males and females. In addition, there was a sex difference in lifestyle-related factors such as exercise, and health examination data such as proteinuria contributing to mortality in middle-aged general inhabitants in rural towns in Japan.