Somatosensory evoked potentials (SEPs) were recorded from the scalp in man to magnetic stimulation of various skeletal muscles. The potentials consisted of several components, the earliest of which decreased in latency as the stimulated site moved rostrally, ranging from 46 msec for stimulation of the gastrocnemius, to 14 msec for stimulation of the deltoid. Experiments were performed to distinguish the mechanisms by which magnetic stimulation of the muscle was effective in evoking these cerebral potentials. For the gastrocnemius, the intensity of the magnetic stimulus needed for evoking cerebral potentials was less than that required for activating mixed or sensory nerves in proximity to the muscle belly (eg, posterior tibial nerve in the popliteal fossa, sural nerve at the ankle). Vibration of the muscle or passive lengthening of the muscle, procedures which activate muscle spindles, were accompanied by a significant attenuation of the potentials evoked by magnetic stimulation of the muscle. Anesthesia of the skin underlying the stimulating coil had no effect on the latency or amplitude of the early components of the magnetically evoked potentials, whereas electrically evoked potentials from skin electrodes were abolished. Thus, the cerebral potentials accompanying magnetic stimulation of the muscle appear to be due to activation of muscle afferents. We suggest that magnetic stimulation of muscle can provide a relatively simple method for quantifying the function of muscle afferents originating from a wide variety of skeletal musculature.