Previous studies have suggested that left ventricular pressure (P) can be predicted in patients with aortic stenosis by the equation P = 235 h/r, where 235 is a constant peak wall stress (sigma), h is end-systolic wall thickness, and r is end-systolic dimension/2; h and r are measured by M-mode echocardiography. In 73 patients with aortic stenosis (valve area less than 0.7 cm2), measured and predicted left ventricular pressure correlated poorly (r = 0.17). The measured wall stress in our patients varied from 120 to 250 mm Hg in patients with normal left ventricular function and from 250 to 550 mm Hg in patients with abnormal function. The correlation between sigma and h was only fair (r = 0.53), because many patients had inappropriate left ventricular hypertrophy. There was a statistically significant correlation between ejection fraction and sigma (r = 0.62) and between ejection fraction and end-systolic dimension (r = -0.70), but there was considerable scatter of ejection fractions for any given end-systolic dimension. We conclude that sigma is not constant in aortic stenosis, and the use of a constant sigma to predict left ventricular pressure is unreliable; inappropriate left ventricular hypertrophy may explain why sigma is not constant. M-mode echocardiography is not reliable in assessing the severity of aortic stenosis in adults; such assessment requires precise measurements of pressure gradients and flow by cardiac catheterization.