To test the applicability of Doppler ultrasound in the evaluation of prosthetic valve function, 107 patients with normal ejection fractions in whom Starr-Edwards, Björk-Shiley, Carpentier-Edwards, and Hancock models had been implanted in the aortic position were examined. Maximal transvalvular velocity was recorded by non-imaging continuous wave Doppler ultrasound. Means of maximal velocities by model and size ranged from less than 2 to 4 m/sec. The Starr-Edwards valve showed the highest velocities, the Björk-Shiley the lowest, and the bioprosthetic models showed velocities in between. A significant inverse relation between velocity and size, and standard deviations averaging +/- 14% enabled the technique to measure differences between sizes of the same model. Aortic regurgitation was detected in 24% of the patients. This study, conducted in well and stable patients, established values for maximal velocity across normally functioning aortic mechanical and tissue prostheses of different models and sizes. The intersubject variability was relatively small which, together with a previously shown minimal intrasubject variability, was testimony to a methodology that should prove useful in longitudinal postoperative evaluations.