Three observers tested which of 2 cues (rate of change in binocular disparity or a comparison of the speeds of motion in the 2 eyes) is used to discriminate the speed of motion-in-depth. In Exp 1, speed discrimination was measured for a dot moving away from the observer in depth (along the z-axis) and for a random dot stereogram in which a central square moved away from the observer in depth. Speed discrimination thresholds were as good for 3-D motion as for monocular sideways motion. In Exp 2, a dynamic random dot stereogram was used to remove consistent monocular cues. 3-D speed discrimination was now very poor, suggesting that the rate of change of disparity is not a good cue for 3-D speed. Exp 3 used a comparison of monocular motion cues to discriminate the speed-in-depth, rather than a single monocular speed cue, suggesting that the visual system contains a binocular system responsive to the speed of motion-in-depth.