It is universally accepted that the postcranial skeleton of the early hominid Australopithecus afarensis shows adaptations, or at least exaptations, towards bipedalism. However, there continues to be a debate concerning the likely form of terrestrial bipedality: whether gait was erect, like our own, or "bent-hip, bent-knee" like the bipedalism of living chimpanzees. In this study we use predictive dynamic modelling to assess the mechanical effectiveness of AL-288-1 under both hypotheses, on the basis of data on segment proportions from the literature. AL-288-1's proportions are incompatible with the kinematics of chimpanzee bipedalism, but compatible with the kinematics of either erect: or "bent-hip, bent-knee" human gait. In the latter case, neither the ankle nor the knee joint would have contributed substantial mechanical work to propulsion of the body, and net energy absorption is predicted for these joints, which would have resulted in increased heat load.. Such an ineffective gait is unlikely to have lead to selection for "bipedal" features in the postcranial skeleton. (C) 1998 Academic Press.