A total of 135 patients with normally functioning prosthetic aortic valves who were catheterized 6 months after placement of Hancock, modified Hancock or Bjork-Shiley prostheses were studied to determine the magnitude of error in Gorlin formula estimates of prosthetic aortic valve area. All patients were male, selected from 13 participating hospitals and routinely followed after valve replacement for 5 years. Hemodynamically determined Gorlin valve areas were compared with independently verified actual valve areas. Actual Hancock areas were measured from videotapes of valves exercised in a pulse duplicator flow model. Actual Bjork-Shiley areas were calculated directly from the valves' inner ring radius. Gorlin valve areas correlated poorly with actual valve areas (r = 0.39). The mean Gorlin formula error was 0.36 cm2 (standard deviation = 0.32). Gorlin areas overestimated actual areas by greater than 0.25 cm2 in 43 patients (32%) and underestimated actual areas by greater than 0.25 cm2 in 29 (21%). It was concluded that the Gorlin formula inaccurately predicts prosthetic valve area in the aortic position. Overreliance on this formula in assessing aortic stenosis could lead to errant clinical decisions.