The exoskeleton is an autonomous robotic device whose function is to increase the strength and endurance of a human pilot. In order to achieve an exoskeleton controller which reacts compliantly to external forces, an accurate model of the dynamics of the system is required. In this report, a series of system identification experiments are designed and carried out for the Berkeley lower extremity exoskeleton. As well as determining the mass and inertia properties of the segments of the legs, various non-ideal elements such as friction, stiffness and damping forces are identified. The resulting dynamic model is found to be significantly more accurate than the original model predicted from the designs of the robot