In populations with higher high-density lipoprotein cholesterol (HDL-C) levels and lower coronary mortality than Western populations, such as in Japan, the beneficial effect of HDL-C on all-cause mortality may be different. Furthermore, prior studies have not focused on very high level of HDL-C. A total of 7175 community Japanese residents without a past history of cardiovascular disease in 300 randomly selected districts were followed for 9.6 years. During follow-up, there were 636 deaths. The multivariate adjusted hazard ratio (HR) of HDL-C for all-cause or cause-specific mortality was calculated using a Cox proportional hazard model adjusted for other cardiovascular risk factors. The all-cause mortality suggested an inverse, graded relation with HDL-C categories; HR for the very high HDL-C category (> or = 1.82 mmol/L), compared with the reference group (1.04-1.55 mmol/L), was 0.73 (95% confidence interval, C.I., 0.50-1.06) for men, 0.63 (95% C.I., 0.41-0.94) for women and 0.70 (95% C.I., 0.53-0.93) when men and women were combined. Serum HDL-C as a continuous variable showed a significant inverse association with all-cause mortality. The cardiovascular mortality indicated a non-significant but inverse graded relation with HDL-C categories. As in the many Western populations, serum HDL-C levels were inversely associated with all-cause mortality in the Japanese general population.