Mortality related to heart rate (HR) increase in the elderly has not yet been well established. To ascertain the relationships among cognitive impairment (CI), mortality, and HR increase, the authors prospectively studied a random sample of elderly subjects stratified according to presence or absence of CI. Elderly subjects randomly selected in 1991 (n = 1332) were followed up for 12 years. Mortality was established in 98.1% of the subjects. When HR was stratified in quartiles (< 69, 70-75, 76-80, and > 80 bpm), mortality was linearly associated with increased HR in all (from 47.7 to 57.0; r2 = .43, p = .019) and in subjects without (from 41.7 to 51.1%; r2 = .50, p = .043) but not in those with CI (from 57.5 to 66.1; r2 = .20, p = .363). Cox regression analysis, adjusted for several variables, shows that HR doesn't predict mortality in all subjects (RR 0.69; 95% CI = 0.27-1.73) or in those with CI (RR 0.91; 95% CI = 0.81-1.02). In contrast, HR predicts mortality in subjects without CI (RR 1.10; 95% CI = 1.00-1.22). Hence, HR increase is a predictor of mortality in elderly subjects without CI. However, when considering all elderly subjects and those with CI, HR increase seems to have no effect on mortality. Thus, CI should be considered when focusing on HR increase as risk factor for mortality in the elderly.