We studied proprioception with the ankle joint and the metacarpophalangeal (MCP) joint of the index finger of humans by use of a method that could distinguish a position sense from a movement sense. The test measured how subjects' ability to detect a fixed displacement of a joint varied with the rate of joint rotation. A position sense should not depend on the speed of joint placement; therefore slow rates of movement should not degrade subjects' ability to sense joint displacements. However, in the absence of a position sense, subjects would presumably rely on movement signals that do depend on the rate of rotation, and their ability to detect displacements should decrease when rate decreases. Subjects could sense small displacements of the ankle (+/- 3.5 degrees) and the MCP joint (+/- 2.5 degrees lateral excursions) with no decrement in performance at speeds as low as 0.25 degrees/min for the ankle and 0.5 degrees/min for the MCP joint (the slowest tested thus far). The findings confirm the existence of a position sense with these joints. Block of the ulnar nerve at the wrist, which paralyzes the interosseous muscles that adduct and abduct the MCP joint but presumably leaves skin and joint mechanisms unaffected, substantially impaired subjects' ability to detect the lateral excursions at slow speeds. Performance fell sharply at speeds less than 128 degrees/min and leveled off at approximately 20% detections at speeds less than 4 degrees/min. Increasing displacement to +/- 7 degrees did not improve performance. Block of the common peroneal nerve at the knee, which paralyzes the ankle dorsiflexor muscles, substantially impaired subjects' ability to detect the +/- 3.5 degrees displacements at slow speeds when the foot was positioned to slacken the plantarflexion muscles (which were not affected by the block). Performance fell sharply at speeds less than 256 degrees/min and approached zero at speeds less than 16 degrees/min. However, positioning the foot to stretch the plantarflexor muscles restored subjects' performance to near normal. Local anesthetic injected into the MCP joint space produced no observable effect on the ability to detect either slow or fast excursions. The joint anesthesia went unnoticed by the subject. We conclude that independent and separable senses exist for limb position and limb movement and that normal position sense requires sensory inputs from the muscles.