Two experiments examined the role of occlusion in reducing perceived multistability in computergenerated point-light walker displays. In Experiment 1, it was found that point-light walker displays lacking occlusion were perceived as multistable and that the addition of occlusion served to reduce this perceived ambiguity. In Experiment 2, it was found that occlusion served two functions in the perception of point-light walker displays: specifying depth order and indicating the presence of implicit occluding forms. These results are discussed in relation to contemporary motion information processing models. All existing models rely solely on the motion and topographical information present in our nonoccluding display; in so doing, these models prove inadequate, since they seek to derive unique perceptions from information seen by observers as being multistable. The findings from the present experiments demonstrate that observers utilize occlusion to reduce multistability, and, thus, rules for processing this information, and/or other forms of constraining information, need to be added to existing motion information processing models.