The present study was designed to develop a new Doppler method to estimate noninvasively the peak-to-peak pressure gradient in patients with aortic stenosis. It was carried out in two steps. In the first study, left heart catheterization data were analysed in 58 patients with aortic stenosis. The peak pressure gradient correlated highly with (r = 0.98) but overestimated significantly the peak-to-peak pressure gradient. The averaged pressure gradient (average of the peak gradient and the gradient at the peak aortic pressure) was found to approximate closely the peak-to-peak pressure gradient (r = 0.99) with a good separation between significant and insignificant aortic stenosis. The ratio of the left ventricular ejection time over the aortic pressure descending time was studied and the beginning of the late one-third systole was chosen as the point for determining the late systolic gradient by the Doppler technique. In the second study, Doppler echocardiography and cardiac catheterization were performed in 35 patients with aortic stenosis. The Doppler-determined peak pressure gradient correlated highly with catheterization-measured peak pressure gradient (r = 0.95) but overestimated significantly the peak-to-peak pressure gradient. The Doppler-determined averaged pressure gradient (average of the peak and the late systolic gradients) estimated accurately the peak-to-peak gradient (r = 0.97) with a good separation between significant and insignificant aortic stenosis. These results demonstrate the limitations of the peak pressure gradient measurement and the reliability of the Doppler-determined averaged pressure gradient for estimation of the peak-to-peak pressure gradient.