The purpose of this study was to compare passive to active testing on the kinematics of the elbow and forearm using a load-controlled testing apparatus that simulates muscle loading. Ten fresh-frozen upper extremities were tested. Active control was achieved by employing computer-controlled pneumatic actuators attached to the tendons of the brachialis, biceps, triceps, brachioradialis and pronator teres. Motion of the radius and ulna relative to the humerus was measured with an electromagnetic tracking system. Active elbow flexion produced more repeatable motion of the radius and ulna than when tested passively (p<0.05). The decrease in variability, as determined from the standard deviation of five successive trials in each specimen, was 76.5 and 58.0% for the varus–valgus and internal–external motions respectively (of the ulna relative to the humerus). The variability in flexion during simulated active forearm supination was 30.6% less than during passive testing. Thus under passive control, in the absence of stability provided by muscular loading across the joint, these uncontrolled motions produce increased variability amongst trials. The smooth and repeatable motions resulting from active control, that probably model more closely the physiologic state, appear to be beneficial in the evaluation of unconstrained kinematics of the intact elbow and forearm.