The predictive power of elevated heart rate for total mortality was evaluated in a Japanese general population. A total of 573 male participants, aged 40 to 64, who underwent a health examination in 1977, were followed until 1994. Heart rate (beats per minute; bpm) was measured using an electrocardiogram. During the 18 years, 82 subjects died; 18 from cerebro-cardiovascular diseases and 36 from cancer. In a multivariate proportional hazards regression model, age, elevated systolic and diastolic blood pressures, antihypertensive medication, heart rate, uric acid, vital capacity (inversely), and serum cholesterol (inversely) were significantly associated with all-cause death. Of these variables, elevated heart rate was the strongest predictor of all-cause death after adjustment for age. Resting heart rate levels were classified into five groups: < 60 (G1), 60-69 (G2), 70-79 (G3), 80-89 (G4), and > or = 90 (G5) bpm. Heart rates of 60-69 (G2) bpm showed the lowest death rate (14.3%) and heart rate > or = 90 (G5) bpm showed the highest death rate (38.2%) after adjustments for age and other confounding factors. The relative risk of G2 versus G5 was 2.68. An increased mortality risk was shown in men whose heart rate was > or = 90 bpm. Moreover, a continuous model suggested a graded increase in risk, so that risk is likely elevated even for heart rates less than 90 bpm, and lowest risk may be around 60 bpm.