1. The involvement of the motor cortex during human walking was evaluated using transcranial magnetic stimulation (TMS) of the motor cortex at a variety of intensities. Recordings of EMG activity in tibialis anterior (TA) and soleus muscles during walking were rectified and averaged. 2. TMS of low intensity (below threshold for a motor-evoked potential, MEP) produced a suppression of ongoing EMG activity during walking. The average latency for this suppression was 40.0 +/- 1.0 ms. At slightly higher intensities of stimulation there was a facilitation of the EMG activity with an average latency of 29.5 +/- 1.0 ms. As the intensity of the stimulation was increased the facilitation increased in size and eventually a MEP was clear in individual sweeps. 3. In three subjects TMS was replaced by electrical stimulation over the motor cortex. Just below MEP threshold there was a clear facilitation at short latency (approximately 28 ms). As the intensity of the electrical stimulation was reduced the size of the facilitation decreased until it eventually disappeared. We did not observe a suppression of the EMG activity similar to that produced by TMS in any of the subjects. 4. The present study demonstrates that motoneuronal activity during walking can be suppressed by activation of intracortical inhibitory circuits. This illustrates for the first time that activity in the motor cortex is directly involved in the control of the muscles during human walking.