Brisk passive flexion of the proximal interphalangeal joint of the middle finger, produced by using a newly devised instrument, elicited evoked potentials on the scalp. The present study carefully excluded the possible contribution of sensory modalities other than proprioception. The initial part of cortical response was a positive deflexion at the contralateral central area (P1 at 34.6 ms after the stimulus). This was followed by a midfrontal negative wave (N1 at 44.8 ms) and a clear positivity at the contralateral centroparietal area (P2 at 48.0 ms). The evoked responses persisted in spite of the abolition of cutaneous and joint afferents of the finger caused by ischemic anesthesia, but they were lost by ischemic anesthesia of the forearm. Thus, the cortical evoked responses obtained in this study most probably reflect muscle afferent inputs. The scalp distribution of P1 suggested that its cortical generator source was different from that of the N20-P20 components of evoked potentials to electrical median nerve stimulation. Brodmann areas 2 and 3a of human brain, which are known to receive deep receptor inputs, are the most plausible generator sites for the early components of the proprioception-related evoked responses. The amplitude of P2 was related to the velocity but not to the magnitude of movement. In conclusion, the present study established a method for recording the evoked responses to the brisk passive movement of the finger joint, which mainly reflect the dynamic aspects of proprioception mediated through muscle afferent.