For patients with stenotic native valves, the modified Bernoulli equation (ΔP = 4v2) may be applied to Doppler-measured transvalvular velocities to yield an accurate estimate of transvalvular gradients. Although it would be useful if the same approach could be used for those with stenotic prosthetic valves, no previous study has validated the Doppler technique in this setting. We therefore recorded simultaneous continuous-wave Doppler flow profiles and transvalvular manometric gradients in 12 catheterized patients in whom all atrial and ventricular pressures were directly measured (transseptal left atrial catheterization and transthoracic ventricular puncture were performed where necessary). A total of 13 prostheses were studied: 11 mitral (seven porcine, three Starr-Edwards, and one Bjiork-Shiley) and two tricuspid (one porcine and one Bjork-Shiley). The Doppler-determined mean gradient was calculated as the mean of the instantaneous gradients (ΔP = 4v2) at 10 msec intervals throughout diastole. The correlation of simultaneous Doppler (DMG) and manometric mean gradients (MG) for the whole group (n 13) demonstrated a highly significant relationship (MG = 1.07DMG ± 0.28; r = .96, p = .0001). The correlation was equally good for porcine valves alone (n = 8) (MG - 0.O6DMG ± 0.55; r = .96, p = .001) and for mechanical valves alone (n = 5) (MG = 1.06DMG -0.04; r = .93, p = .02). In a subset of patients without regurgitation (n = 8), prosthetic valve areas were estimated by two Doppler methods originally described by Holen and Hatle, as well as by the invasive Gorlin method. As expected from theoretical considerations, a close correlation was not demonstrated between results of the Gorlin method and those of either Hatle's Doppler method (r = .65, fp = NS) or Holen's method (r = .14, p - NS). Comparison of the results of the two Doppler methods yielded a somewhat closer correlation (r = .73, p .05). These results suggest that in patients with disk-occluder, ball-occluder, and porcine prosthetic valves, Doppler estimates of transvalvular gradients are virtually identical to those obtained invasively.