During the last decade deep brain stimulation (DBS) has become a routine method for the treatment of advanced Parkinson's disease (PD), leading to striking improvements in motor function and quality of life of PD patients. It is associated with minimal morbidity. The rationale of targeting specific structures within basal ganglia such as the subthalamic nucleus (STN) or the internal segment of the globus pallidus (GPi) is strongly supported by the current knowledge of the basal ganglia pathophysiology, which is derived from extensive experimental work and which provides the theoretical basis for surgical therapy in PD. In particular, the STN has advanced to the worldwide most used target for DBS in the treatment of PD, due to the marked improvement of all cardinal symptoms of the disease. Moreover on-period dyskinesias are reduced in parallel with a marked reduction of the equivalent daily levodopa dose following STN-DBS. The success of the therapy largely depends on the selection of the appropriate candidate patients and on the precise implantation of the stimulation electrode, which necessitates careful imaging-based pre-targeting and extensive electrophysiological exploration of the target area. Despite the clinical success of the therapy, the fundamental mechanisms of high-frequency stimulation are still not fully elucidated. There is a large amount of evidence from experimental and clinical data that stimulation frequency represents a key factor with respect to clinical effect of DBS. Interestingly, high-frequency stimulation mimics the functional effects of ablation in various brain structures. The main hypotheses for the mechanism of high-frequency stimulation are: (1) depolarization blocking of neuronal transmission through inactivation of voltage dependent ion-channels, (2) jamming of information by imposing an efferent stimulation-driven high-frequency pattern, (3) synaptic inhibition by stimulation of inhibitory afferents to the target nucleus, (4) synaptic failure by stimulation-induced neurotransmitter depletion. As the hyperactivity of the STN is considered a functional hallmark of PD and as there is experimental evidence for STN-mediated glutamatergic excitotoxicity on neurons of the substantia nigra pars compacta (SNc), STN-DBS might reduce glutamatergic drive, leading to neuroprotection. Further studies will be needed to elucidate if STN-DBS indeed provides a slow-down of disease progression.