The Framingham study has shown by M-mode echocardiography that left ventricular hypertrophy is a powerful, independent predictor for the development of coronary heart disease and that increased left atrial dimension has been associated with an increased risk of stroke. No previous population-based study has evaluated the risk factor correlates and predictive value for coronary heart disease and stroke of two-dimensional and Doppler, as well as M-mode, echocardiography. The Cardiovascular Health Study is a multi-year prospective epidemiologic study of 5201 men and women older than 65 recruited from four geographic sites in the United States. The main objectives of incorporating echocardiography were to determine whether echocardiographic indices, or changes in these indices, are (1) correlated with traditional risk factors for coronary heart disease and stroke; and (2) independent predictors of morbidity and mortality for coronary heart disease and stroke. Echocardiographic measurements of interest include those related to global and segmental left ventricular systolic and diastolic structure and function and left atrial size. For each subject, a baseline echocardiogram was recorded in super-VHS tape using a standard protocol and equipment. All studies were sent to a reading center where images were digitized and measurements were made using customized computer algorithms. Calculated data and images were stored on optical disks to facilitate retrieval and future comparisons in longitudinal studies. A second echocardiogram is scheduled in year 7, with a goal of determining whether changes in cardiac anatomy or function over a 5-year period are important predictors of morbidity or mortality from coronary heart disease and stroke. Quality control measures included standardized training of echocardiography technicians and readers, technician observation by a trained echocardiographer, periodic blind duplicate readings with reader review sessions, phantom studies, and quality control adults.