From the cohort taking part in the Atherosclerosis Risk in Communities (ARIC) study, a multicenter investigation of atherosclerosis and its sequelae in women and men ages 45-64 years, a sample of 145 subjects with significant carotid artery atherosclerosis but without clinically recognized coronary heart disease was identified along with 224 group-matched control subjects. The aim of this paper is to measure the association of the apolipoprotein (apo) E polymorphism with the prevalence of significant carotid artery atherosclerotic disease (CAAD) after considering the contribution of established risk factor variables. The first model used a stepwise selection procedure to define a group of significant physical and lifestyle characteristics and a group of significant plasma lipid, lipoprotein, and apolipoprotein variables that were predictive of CAAD status in this sample. Those variables selected included age (years), body mass index (BMI; kg/m2), consumption of cigarettes (CigYears; number of cigarettes/d x the number of smoking years), hypertension status, high-density lipoprotein (HDL)-cholesterol (mg/dl), total cholesterol (mg/dl), and Lp[a] (micrograms/ml). The second model was built by forcing into the equation an a priori set of demographic, anthropometric, and lipoprotein variables, which were age, BMI, CigYears, hypertensive status, LDL-cholesterol, and HDL-cholesterol. In both models, the apo E genotype epsilon 2/3 was related to CAAD status. For both models, the estimated odds ratio of being a CAAD case associated with the apo E genotype epsilon 2/3 was > 2:1. The mechanism of the observed association between the epsilon 2/3 genotype and carotid atherosclerosis is unknown, but it is likely due to the known effects of the E2 isoform in causing delayed clearance of triglyceride-rich lipoproteins.