The aim of the study was to determine whether the integration of dynamic weight shifting into treadmill training would improve the efficacy of treadmill training in humans with spinal cord injury. DESIGN: Sixteen humans with spinal cord injury were randomly assigned to receive robotic or treadmill-only training and underwent 6 wks of training. A force was applied to the pelvis for facilitating weight shifting and to the legs for assisting with leg swing for participants in the robotic group. No assistance force was applied for participants in the treadmill-only group. Outcome measures consisted of overground walking speed, 6-min walking distance, and other clinical measures and were assessed before, after 6 wks of training, and 8 wks after the end of training. RESULTS: A greater improvement in 6-min walking distance was observed after robotic training than that after treadmill-only training (P = 0.03), but there was not a significant difference between the two groups in improvements in walking speed. However, a greater improvement was observed for the participants who underwent robotic training than those who underwent treadmill-only training (i.e., 15% vs. 2%). CONCLUSIONS: Applying a pelvis assistance force for facilitating weight shifting during treadmill training may improve locomotor function in humans with spinal cord injury.