Although most tasks performed by human observers that require accurate stereoscopic depth perception, such as working with tools, operating machinery, and controlling vehicles, involve dynamically changing disparities, classification of observers as having normal or deficient stereoscopic vision is currently based on performance with static stereoscopic displays. The present study compares the performance of subjects classified as deficient in static stereoscopic vision to a control group with normal stereoscopic vision in two experiments-one in which the disparities were constant during motion and one in which the disparities changed continuously. In the first experiment, subjects judged orientation in depth of a dihedral angle, with the apex pointed toward or away from them. The angle translated horizontally, leaving the disparities constant. When disparity and motion parallax were placed in conflict, subjects in the normal group almost always responded in accordance with disparity, whereas subjects in the deficient group responded in accordance with disparity at chance levels. In the second experiment, subjects were asked to judge the direction of rotation of a computer-generated cylinder. When dynamic occlusion and dynamic disparity indicated conflicting directions, performance of subjects in the normal and deficient groups did not differ significantly. When only dynamic disparity information was provided, most subjects classified as stereo deficient were able to judge the direction of rotation accurately. These results indicate that measures of stereoscopic vision that do not include changing disparities may not provide a complete evaluation of the ability of a human observer to perceive depth on the basis of disparity.