Cortex-muscle coherence is a frequency-analysis technique that has been increasingly applied in the investigation of movement disorders. To study the intra- and inter-session stability of the cortex-muscle coherence, we recorded from 12 healthy subjects magnetoencephalographic (MEG) and surface electromyographic (EMG) signals during unilateral isometric contractions of the left- and right-hand muscles. Two identical measurements were performed during one session, and the session was repeated once after about 1 year. In one experienced subject, the recordings were repeated seven times within 20 months. The MEG-EMG coherence exceeded the noise level in 10 out of 12 subjects. Both the frequency (correlation coefficient r = 0.77-0.93, P < 0.01) and strength (r = 0.78-0.91, P < 0.01) of coherence were well reproducible within each session for both left- and right-sided contractions. The inter-session reproducibility was high for the mean of cumulative coherence frequency (r = 0.90-0.95, P < 0.01), but relatively low for coherence strength (r = 0.43-0.59, P > 0.05). The results for one subject participating in 8 repeated sessions strongly supported the results of the whole group. Thus, intra-session reproducibility of both strength and frequency of the cortex-muscle coherence is good and studies comparing different conditions at the group level within one session are feasible. However, caution is needed when interpreting absolute levels or changes in the strength of coherence in single subjects between the sessions.