Some of the most long-standing questions in paleoanthropology concern how and why human bipedalism evolved. Over the last century, many hypotheses have been offered on the mode of locomotion from which bipedalism originated. Candidate ancestral adaptations include monkey-like arboreal or terrestrial quadrupedalism, gibbon- or orangutan-like (or other forms of) climbing and suspension, and knuckle-walking. This paper reviews the history of these hypotheses, outlines their predictions, and assesses them in light of current phylogenetic, comparative anatomical, and fossil evidence. The functional significance of characteristics of the shoulder and arm, elbow, wrist, and hand shared by African apes and humans, including their fossil relatives, most strongly supports the knuckle-walking hypothesis, which reconstructs the ancestor as being adapted to knuckle-walking and arboreal climbing. Future fossil discoveries, and a clear understanding of anthropoid locomotor anatomy, are required to ultimately test these hypotheses. If knuckle-walking was an important component of the behavioral repertoire of the prebipedal human ancestor, then we can reject scenarios on the origin of bipedalism that rely on a strictly arboreal ancestor. Moreover, paleoenvironmental data associated with the earliest hominins, and their close relatives, contradict hypotheses that place the agents of selection for bipedality in open savanna habitats. Existing hypotheses must explain why bipedalism would evolve from an ancestor that was already partly terrestrial. Many food acquisition and carrying hypotheses remain tenable in light of current evidence. Yrbk Phys Anthropol 44:70–105, 2001. © 2001 Wiley-Liss, Inc.