To investigate the distribution of heart rate variability (HRV) spectral power in an unselected sample of the population, and to ascertain the population correlates of HRV, we examined 1,984 healthy persons, aged 45 to 64 years, randomly selected from the Atherosclerosis Risk in Communities (ARIC) study cohort. Resting, supine, 2-minute, beat-to-beat heart rate data were collected between 7 A.M. and 12 noon. The race- and sex-adjusted geometric means of low-frequency component (LF, 0.025 to 0.15 Hz) were 4.00 and 3.13 (beats/min)2; of high-frequency component (HF, 0.16 to 0.35 Hz), 1.65 and 1.21 (beats/min)2; and of the HF/LF ratio, 0.41 and 0.39, for 45-to-54 and 55-to-64 years age groups, respectively (test of mean difference by age, p < 0.01, p < 0.01, and p = 0.11 for LF, HF, and HF/LF ratio, respectively). Comparing black with white examinees, the age- and sex-adjusted geometric means of LF were 3.06 and 3.70 (beats/min)2; of HF, 1.66 and 1.36 (beats/min)2; of HF/LF, 0.54 and 0.37, respectively (test of mean difference by race, p < 0.01, p < 0.01, and p < 0.01). The age- and race-adjusted geometric means of LF for women and men were 3.12 and 4.10 (beats/min)2; of HF, 1.46 and 1.38 (beats/min)2; and of HF/LF, 0.47 and 0.34, respectively (test of mean difference, p < 0.01, p = 0.34, and p < 0.01). We conclude that HRV spectral indexes are associated with age, race, and sex. With increasing age, the parasympathetic and sympathetic spectral power components decrease.(ABSTRACT TRUNCATED AT 250 WORDS)