Echocardiographic measures of left ventricular (LV) function and structure as well as left atrial size have been reported to predict adverse cardiovascular disease (CVD) outcomes. Although anatomic changes of the aortic root are likely to reflect effects of hypertension and atherosclerosis, few data are available on the predictive value of aortic root dimension (ARD) for outcome in free-living populations. The purpose of this investigation was to determine whether in a cohort of patients aged > or = 65 years ARD was associated with traditional coronary heart disease (CHD) risk factors and with 10-year incident CVD outcomes. In the National Heart, Lung, and Blood Institute-sponsored Cardiovascular Health Study, 3,933 free-living black and white men and women > or = 65 years of age without prevalent CVD had 2-dimensional directed M-mode echocardiographic measurements of ARD as part of a comprehensive evaluation. ARD was associated with age and gender (greater in men) but not race. ARD was also positively associated with diastolic blood pressure, LV hypertrophy, major electrocardiographic abnormalities, and other echocardiographic measures, including LV mass, ventricular septal and posterior wall thickness, and LV dimension. After adjustment for other known risk factors, high ARD was associated with an increased risk for incident congestive heart failure (CHF) in men (hazard ratio for upper compared with all other quintiles 1.47, p = 0.014), stroke in men and women (hazard ratio 1.39 per cm, p = 0.015), CVD mortality in men and women (hazard ratio 1.48 per cm, p = 0.007), and total mortality in men and women taking antihypertensive medications (hazard ratio 1.46 per cm, p = 0.007), but not with incident myocardial infarction (MI) (hazard ratio 0.89, p = 0.39). In conclusion, in a cohort of patients aged > or = 65 years without clinical CVD at baseline, ARD was associated with several CHD risk factors and measures of subclinical disease and was predictive of incident CHF, stroke, CVD mortality, and all-cause mortality, but not of incident MI.