Twelve patients with deafferentation pain secondary to central nervous system lesions were subjected to chronic motor cortex stimulation. The motor cortex was mapped as carefully as possible and the electrode was placed in the region where muscle twitch of painful area can be observed with the lowest threshold. 5 of the 12 patients reported complete absence of previous pain with intermittent stimulation at 1 year following the initiation of this therapy. Improvements in hemiparesis was also observed in most of these patients. The pain of these patients was typically barbiturate-sensitive and morphine-resistant. Another 3 patients had some degree of residual pain but considerable reduction of pain was still obtained by stimulation. Thus, 8 of the 12 patients (67%) had continued effect of this therapy after 1 year. In 3 patients, revisions of the electrode placement were needed because stimulation became incapable of inducing muscle twitch even with higher stimulation intensity. The effect of stimulation on pain and capability of producing muscle twitch disappeared simultaneously in these cases and the effect reappeared after the revisions, indicating that appropriate stimulation of the motor cortex is definitely necessary for obtaining satisfactory pain control in these patients. None of the patients subjected to this therapy developed neither observable nor electroencephalographic seizure activity.