One hundred and nineteen patients with a clinical diagnosis of important aortic stenosis were assessed clinically and by electrocardiography and M mode echocardiography to determine the degree of left ventricular hypertrophy. Predicted left ventricular pressure was calculated from two previously described formulas. Comparisons were made between the various methods for assessing left ventricular hypertrophy to see which method most reliably predicted the severity of the stenosis as defined by invasive measurement of left ventricular pressure and peak aortic valve gradient. Direct measurement of left ventricular wall thickness from the echocardiogram, expressed as the mean of septal and posterior wall thickness, was the most accurate predictor (r = 0.75 for 29 patients with high quality echocardiograms), and surpassed derived indices (left ventricular mass (r = 0.68) and predicted left ventricular pressure derived from the two formulas (r = 0.39 and r = 0.68)) in adults. Echocardiographic results were significantly better than electrocardiographic, but only when the recordings were of very high quality. Average quality echocardiograms were no better than precordial electrocardiographic voltages for predicting the severity of aortic stenosis. The formulas for predicting left ventricular pressure were of more value in children than in adults, but they were still not sufficiently accurate to be predictive in individual cases. Electrocardiographic voltages were more accurate predictors of the severity of aortic stenosis in children than in adults.