The present renewal of the surgical treatment of Parkinson's disease, almost abandoned for twenty Years, arises from two main reasons. The first is the better understanding of the functional organization of the basal ganglia. It was demonstrated in animal models of Parkinson's disease that the loss of dopaminergic neurons within the substantia nigra, at the origin of the striatal dopaminergic defect, induces an overactivity of the excitatory glutamatergic subthalamo-internal pallidum pathway. The decrease in this hyperactivity might lead to an improvement in the pakinsonian symptoms. The second reason is the improvement in stereotactic neurosurgery in relation with the progress in neuroimaging techniques and with intraoperative electrophysiological microrecordings and stimulations, which help determine the location of the deep brain targets. In the 1970s chronic deep brain stimulation in humans was applied to the sensory nucleus of the thalamus for the treatment of intractable pain. In 1987, Benabid and colleagues suggested high frequency stimulation of the ventral intermediate nucleus of the thalamus in order to treat drug-resistant tremors and to avoid the adverse effects of thalamotomies. How deep brain stimulation works is not well known but it has been hypothetized that it could change the neuronal activities and thus avoid disease-related abnormal neuronal discharges. Potential candidates for deep brain stimulation are selected according to exclusion and inclusion criteria. Surgery can be applied to patients in good general and mental health, neither depressive nor demented and who are severely disabled despite all available drug therapies but still responsive to levodopa. The first session of surgery consists in the location of the target by ventriculography and/or brain MRI. The electrodes are implanted during the second session. The last session consists in the implantation of the neurostimulator. The ventral intermediate nucleus of the thalamus was the first target in which chronic deep brain stimulation electrodes were implanted in order to alleviate tremor. This technique can be applied bilaterally without the adverse effects of bilateral thalamotomies. Like pallidotomy, internal globus pallidum stimulation has a dramatic beneficial effect on levodopa-induced dyskinesia but its effects on the parkinsonian triad are less constant and opposite motor effects are sometimes observed in relation with the stimulated contact. The inconstant results, perhaps related to the complexity of the structure led to the development of subthalamic nucleus stimulation. The alleviation of motor fluctuations and the improvement in all motor symptoms allows a significant decrease in levodopa daily dose and in levodopa-induced dyskinesia. Presently, deep brain stimulation is a fashionable neurosurgical technique to treat Parkinson's disease. Subthalamic nucleus stimulation seems to be the most suitable target to control the parkinsonian triad and the motor fluctuations. Because of the possible adverse effects it must be reserved for disabled parkinsonian patients. No large randomized study comparing different targets and different neurosurgical techniques has been performed yet. Such studies, including cost benefit studies would be useful to assess the respective value of these different techniques.