Single motor units were isolated in medial gastrocnemius (MG) or soleus (Sol) muscles of the cat. Single shocks delivered to the motor axon elicited EMG waveforms which were recorded in the muscle. The amplitude of each individual EMG waveform as well as the area under the full-wave rectified waveform were measured. Mechanical properties of the motor units were then measured so that each unit could be classified as FF, FR or S according to the criteria of Burke. Finally, each unit was stimulated continuously at 80 pulses per second (pps), and changes in electrical and mechanical responses were recorded and compared. IEMG was positively correlated with maximum tetanic tension of motor units in MG. The relationship could be fitted with a parabola showing that 'large' motor units produce relatively more electrical activity than do 'small' motor units. Two types of electrical changes were seen during continuous stimulation of motor units at 80 pps: a smooth continuous decline in IEMG was attributed to electrical changes occurring at a site or sites distal to the neuromuscular junction; and random, abrupt changes in unit-EMG waveform and hence IEMG amplitude, were attributed to failure of the action potential to propagate past axonal branch points, resulting in the intermittent failure of groups of muscle fibers to respond to stimuli. The rates of decline of electrical and mechanical activity were compared for motor units. It was found that in fast fatiguing motor units, tetanic tension declined more rapidly than did IEMG, while in slowly fatiguing motor units, IEMG declined more rapidly than did force. We conclude that fatigue of motor units induced by continuous 80 pps stimulation can occur at different sites in the motor unit, and that the site of fatigue depends on motor unit properties or motor unit type.