We measured the body sway of test subjects who were viewing stereoscopic images. In order to simulate a real plate moving in depth, the binocular disparity and the visual angle of the images were varied. In the case of varying binocular disparity, all observers could not perceive motion in depth when watching the moving image, but body sway was still observed. For all observers the local peaks of the power spectrum data were observed at the frequencies of the image motion. In the case of varying visual angle, motion in depth was always perceived, but did not produce body sway. This suggests that the cue to motion in depth is different from the cue to body sway.