To determine whether motor evoked potential (MEP) amplitude and area are accurate measurements of the magnitude of response to magnetic cortical stimulation, we simultaneously recorded the twitch and MEP in the first dorsal interosseous muscle of 8 normal subjects. Consecutive stimuli were delivered at increasing stimulus intensities (SI) or with increasing levels of background voluntary muscle contraction (BVC). There was stimulus to stimulus variability in MEP amplitude, area and twitch force. At low SI and at low levels of background contraction, there was a good correlation between twitch amplitude and MEP amplitude and area (r = 0.6-0.96, P < 0.005). Increasing either variable caused the correlation to decrease significantly (r = 0.02-0.31, P > 0.01). With increasing SI, MEP amplitude and area plateaued but twitch force continued to increase. A similar pattern was observed with higher levels of background muscle contraction although in some subjects a second increase in MEP amplitude and area was seen. Collision experiments demonstrated that the amplitude of the EMG activity resulting from repetitive motoneuron firing increased as SI was increased. This is due to multiple descending volleys which result in repetitive firing of some spinal motoneurons. Rapid, repetitive firing of some motor units is likely to result in phase cancellation and, therefore, the MEP amplitude, and to a lesser extent area, do not accurately reflect the net motor output.