To allow for efficient and robust gait pattern, humans instantaneously modulate their joint impedance. Lower-limb amputees and patients suffering from neurological diseases partly lack that ability. Therefore, prosthetic and rehabilitative devices have been designed to re- store and repair nominal locomotion. For bio-inspired control of these devices, the underlying physiological behavior of the lower-extremity must be identified and quantified. Deeper under- standing is also important to determine appropriate therapy for patients with upper motor neuron diseases. Different identification methods exist to quantify joint impedance during well-controlled, static tasks using continuous random perturbations. However, methods are lacking that which quantify joint impedance during walking The goal of this Master thesis was to experimentally validate a novel method to identify joint impedance during the stance phase of walking. The thesis investigated whether transient, endpoint perturbations applied using an instrumented treadmill were sufficient to identify the dynamics of a single-joint system where all properties were known. The influences from the experimental setup were evaluated and finally, the method was applied in a pilot study of a human subject standing on the treadmill. Results indicate that joint parameters could be consistently estimated with a good fit. How- ever, they were not physically meaningful and did not match the true parameters of the system. This was caused by influences from the experimental setup, which could not be di- rectly subtracted from the data. Model simulation demonstrated the sensitivity of the method to measurement noise and that the set of estimated parameters were not unique. The limitations of the method and the available experimental setup were carefully identified in this study. Thorough guideline for the method was developed which facilitates the use of the method and sharpens the goal for further research.