Body-weight supported treadmill training following neurological disorders such as stroke and spinal cord injuries (SCI) has become an integral part of rehabilitation for treating gait disorders. Unfortunately techniques for selecting important training parameters, such as walking speed and body-weight support, have not been established. Here we present a 3-D inverse-dynamics based approach for evaluating an individual's ability to ambulate, in terms of evaluating the magnitude and timing of joint moments at the ankle, knee and hip. This technique, which utilizes an instrumented gait robot, allows clinicians and researchers the ability to determine the training parameters in which subjects generate joint moments at the proper phases of the gait cycle which when combined with electromyographic recordings, can help establish and then progress training parameters for individuals on a subject-by-subject basis. We believe that training subjects at their preferred walking speed and body-weight support will lead to higher functional outcomes.