To test the hypothesis that subtle changes in the resting electrocardiogram can have predictive value for subsequent cardiovascular disease, the authors evaluated serial electrocardiograms from a cohort of initially healthy men and related these to later development of ischemic heart disease. The cohort of 1056 men originally physically qualified for naval aviation were followed from 24-61 years of age between 1940 and 1977. Resting electrocardiograms were obtained at 24, 36, 42, and 54 years of age. Particular emphasis was directed to measurement of electrocardiographic intervals, amplitudes, and vectorial orientation. The potential effects of weight, blood pressure, and other major risk factors on electrocardiographic changes were controlled in analysis. At younger ages, the amplitude of the T wave in lead 2 and the change in P-R interval were predictive of cardiovascular disease. With aging of the cohort, more leftward orientation of the frontal plane QRS and T forces were predictive, as were the serial changes in the angle between QRS and T. When major risk factors (smoking, blood pressures, and serum cholesterol) were included in a predictive model, addition of electrocardiographic changes significantly improved prediction of ischemic heart disease. These subtle serial changes may be helpful to epidemiologists and clinical investigators in assessing risk of subsequent cardiovascular disease.