This study compares the cross-sectional and longitudinal associations of cardiovascular risk factors with clinical coronary heart disease (CHD) and with subclinical carotid atherosclerosis measured by ultrasound. The study population were 1410 participants in the Atherosclerotic Risk in Community (ARIC) Study (1987-1989) who also participated in a 1974 community health survey. Smoking in 1974 was associated with increased CHD prevalence in 1987-1989 (adjusted prevalence ratio = 2.2), whereas the corresponding cross-sectional association was practically absent. For hypercholesterolemia and hypertension, the longitudinal associations with CHD were also stronger than the cross-sectional associations. In contrast, the strength of the longitudinal and cross-sectional associations with carotid atherosclerosis was generally similar. These results underscore the advantages of using subclinical measures of atherosclerosis in cross-sectional studies. In addition, they suggest that the presence of smoking, hypertension, or hypercholesterolemia in mid-adulthood may have some persisting effects on the development of atherosclerotic disease in later life.