We reviewed prospective epidemiological data in the general population, mostly middle-aged to older persons, to determine the association of carotid intima-media thickness (CIMT) (assessed by B-mode ultrasonography) with cardiovascular risk. Reported risks were expressed as absolute (event risk per persons-years in subjects with a high CIMT) and relative (hazard ratio of high vs low CIMT). They were hardly comparable as the result of differences between the analyzed studies, including the site and procedure of CIMT measurement, the report of adjusted or unadjusted models, and the arbitrary cutoff point to evaluate the CIMTAEs ability to predict risk. Despite these heterogeneities, the following four main conclusions emerged: (1) CIMT was an independent but relatively modest (as judged by absolute risk) predictor of coronary heart disease (CHD); (2) CIMT was an independent predictor for stroke, slightly better than for CHD as judged by the relative risks of both events; (3) CIMT added little to the CHD prediction by risk factors, as judged by c statistic and receiver operating characteristic curve analysis (however, appropriate data for stroke on this important issue were lacking); and (4) the CHD prediction by CIMT was inferior to that by ultrasonography-assessed carotid plaque because plaque may be more representative of atherosclerosis than CIMT.