The purpose of this study was to use a canine preparation of experimental aortic stenosis to compare estimates of pressure gradient derived from continuous wave Doppler ultrasound with gradients measured directly by catheterization. Aortic stenosis was created in six mongrel dogs by placing an elastic band around the aorta. Eighty-eight different pressure gradients, ranging from 5 to 160 mm Hg, were produced by variable tightening of the aortic band. Pressure gradients were measured by micromanometer-tipped catheters placed in the left ventricle and aorta. Doppler spectral signals were simultaneously obtained using a 2.0 MHz nonimaging transducer placed directly on the surface of the ascending aorta. Doppler and pressure recordings were analyzed using a custom-designed software program to measure maximal instantaneous, mean and peak to peak gradients, as well as ejection and acceleration times. Maximal instantaneous Doppler gradient showed an excellent linear correlation with maximal instantaneous catheterization gradient (r = 0.98, SEE = 5.3 mm Hg). The correlation of Doppler-estimated maximal gradient to peak to peak catheterization gradient was also linear (r = 0.97, SEE = 6.2 mm Hg) but resulted in a systematic overestimation of pressure drop (mean overestimation = 9.0 mm Hg). Measurement of the Doppler gradient at mid-systole resulted in a more accurate correlation with the peak to peak catheterization gradient (r = 0.98, SEE = 6.1 mm Hg) and eliminated the problem of overestimation.(ABSTRACT TRUNCATED AT 250 WORDS)