To evaluate the normal range of Doppler-derived velocities and gradients, their relation to direct flow measurements and the importance of prosthetic valve design on the relation between Doppler and catheter-derived gradients, five sizes of normal St. Jude bileaflet, Medtronic-Hall tilting disc, Starr-Edwards caged ball and Hancock bioprosthetic aortic valves were studied with use of a pulsatile flow model. A strong linear correlation between peak velocity and peak flow, and mean velocity and mean flow, was found in all four valve types (r = 0.96 to 0.99). In small St. Jude and Hancock valves, Doppler velocities and corresponding gradients increased dramatically with increasing flow, resulting in velocities and gradients as high as 4.7 m/s and 89 mm Hg, respectively. The ratio of velocity across the valve to velocity in front of the valve (velocity ratio) was independent of flow in all St. Jude, Medtronic-Hall, Starr-Edwards and Hancock valves when the two lowest flow rates were excluded for Hancock valves. Although Doppler peak and mean gradients correlated well with catheter peak and mean gradients in all four valve types, the actual agreement between the two techniques was acceptable only in Hancock and Medtronic-Hall valves. For St. Jude and Starr-Edwards valves, Doppler gradients significantly and consistently exceeded catheter gradients with differences as great as 44 mm Hg. Thus, Doppler velocities and gradients across normal prosthetic heart valves are highly flow dependent. However, the velocity ratio is independent of flow.(ABSTRACT TRUNCATED AT 250 WORDS)