A 62-year-old man was injured in a road-traffic accident 8 years before and his right lower leg was amputated later 1! s years later because of necrotic changes. After amputation, he had phantom limb pain, described as continuous burning and intermittent severe crushing pain. He could not sit still for even 3 min because of pain. He was admitted to the pain clinic and treated with drugs, lumbar sympathetic ganglion block, and dorsal column stimulation but did not get relief from his pain. He was admitted to the department of neurosurgery for motor cortex stimulation. The location of the precentral gyrus was estimated from magnetic resonance images and angiography. On December 10, 1997, an electrode array with four plate electrodes 0·5 cm in diameter, each separated by 0·5 cm (Model 3586, Medtronic, Inc, Minneapolis, Minnesota), was placed on the left interhemispheric motor cortex under general anaesthesia. Test stimulation showed that a visual analogue scale of his pain was reduced from 9 to 2, and that the pain reduction lasted for a day. On Dec 17, 1997, the stimulation system was internalised (XTREL, Medtronic). At follow-up in December, 1998, pain reduction with stimulation is still effective. The mechanism of phantom limb pain is not known; however, both hyperactivity of peripheral nerves and sensitisation of spinal neurons may play a part. Experimental studies have shown that some deafferented dorsal root ganglion cells developed spontaneous activity after dorsal rhizotomy and that prolonged or repetitive input from unmyelinated afferent nociceptors produce progressive increase of neuronal responses in the dorsal horn to subsequent peripheral stimuli. On the other hand, Melzack proposed that the anatomical structure of the body was represented in a neuromatrix extending throughout the brain which is genetically determined and later modified by sensory inputs. Nerve deafferentation can eventually result in neuron damage and pain generated in the brain. Pharmacological treatments of phantom limb pain with tricyclic antidepressants, carbamazepine, and ketamine are effective in some cases. Sympathetic ganglion blockade may work in patients with burning pain. Dorsal column stimulation provides a third of patients with a greater than 50% reduction in pain. Excisions of cortical areas or of the thalamus are effective for a short time, after which pain returns. There are two general targets in the brain (periaqueductal grey and thalamus) for electrical stimulation to treat pain. Only thalamic stimulation has sometimes brought pain relief in deafferentation pain. Motor cortex stimulation is less invasive and effective for approximately half of the patients with central post-stroke deafferentation pain. It was thought that activation of hypothetical sensory neurons through motor cortex stimulation may inhibit deafferentation nociceptive neurons within the cortex.