The recovery of locomotion, following interactive training with graded weight support, in the adult spinal cat has led to the proposal that removal of body weight may be a therapeutic tool in human gait retraining. There would be benefits, however, in knowing normal responses of humans to partial weight bearing before applying this strategy to patients. In this study, 10 nondisabled male subjects walked on a treadmill while 0%, 30%, 50%, and 70% of their body weight was supported by a modified climbing harness. To dissociate the changes attributable to walking speed from those attributable to body weight, each subject walked at the specified body-weight-support (BWS) levels and at full weight bearing (FWB) at the same speed. Simultaneously, electromyographic data from the right leg muscles, footswitch signals, and video recording of joint motion were collected. The FWB and BWS gaits appeared similar, except at the highest level of BWS studied (ie, 70% of BWS). Significant differences among other BWS and FWB trials at comparable speeds included decreases in percentage of stance, percentage of total double-limb support time, and maximum hip and knee flexor swing angle. Other adaptations to BWS were a reduction in the mean burst amplitude of the muscles that are active during stance and an increase in the mean burst amplitude of the tibialis anterior muscle. The possible implications of this new gait retraining strategy for patients with neurological impairment are discussed. [Finch L, Barbeau H, Arsenault B. Influence of body weight support on normal human gait: development of a gait retraining strategy.