Background. Data from the Atherosclerosis Risk in Communities (ARIC) cohort study were examined both cross-sectionally and intraindividually to confirm recent findings from population-based studies showing a decline in total cholesterol (TC) levels in the United States. Methods. For the cross- sectional analysis, mean plasma TC levels from 15,792 participants aged 45-64 at baseline visit, and who were selected randomly from four U.S. communities, were examined for each year covered by the first cohort visit (1987, 1988, and 1989). Ninety-three percent of the cohort participants returned for the follow-up visit (1990, 1991, and 1992), and were included in the assessment of intraindividual TC trends. Results. Both mean TC and prevalence of hypercholesterolemia (defined as plasma cholesterol concentration ≥240 mg/dl) consistently declined over the 3 years covered by visit 1 for all age- gender-race groups. For 1987, 1988, and 1989, mean TC values (mg/dl) were, respectively, 220.3, 216.7, and 214.1 (annual average change, -1.4%, P < 0.001). For these same years, hypercholesterolemia prevalence rates were 30.0, 27.8, and 25.3% (annual average change, -7.8%, P<0.001). The mean plasma TC also decreased within individuals between the two visits across race, gender, and age decade categories. With the exception of black men, this decline was more marked for older than younger subjects, but no consistent differences were seen between the racial groups. However, in whites, decreases were greater for men than for women. Expected results were seen when these changes were correlated with changes in cardiovascular risk factors between the two visits. Conclusion. The current study results are consistent with those of previous studies, and confirm the notion that preventive programs appear to be effective in reducing mean population TC levels. (C) 2000 American Health Foundation and Academic Press.