Dystonia has commonly been thought to represent a disorder of basal ganglia function. Although long considered a hyperkinetic movement disorder, the evidence to support such a classification was based on the presence of excessive involuntary movement, not on physiological data. Only recently, with the return of surgical procedures using microelectrode guidance for the treatment of dystonia, has electrophysiological data demonstrated an alteration in mean discharge rate, somatosensory responsiveness and the pattern of neuronal activity in the basal ganglia thalamocortical motor circuit. Previous models of dystonia suggested that reduced mean discharge rates in the globus pallidus internus (GPi) led to unopposed increases in activity in the thalamocortical circuit that precipitated the development of involuntary movement associated with dystonia. This model has subsequently been modified given the clear improvement in dystonic symptoms following lesions in the GPi, a procedure that is associated with a further reduction in pallidal output. The improvement in dystonia following pallidal lesions is difficult to reconcile with the "rate" hypothesis for hypokinetic and hyperkinetic movement disorders and has led to the development of alternative models that, in addition to rate, incorporate changes in pattern, somatosensory responsiveness and degree of synchronization of neuronal activity. Present models of dystonia, however, must not only take these changes into account but must reconcile these changes with the reported changes in cortical excitability reported with transcranial magnetic stimulation, the changes in metabolic activity in cortical and subcortical structures documented by positron emission tomography (PET), and the alterations in spinal and brainstem reflexes. A model incorporating these changes together with the reported changes in neuronal activity in the basal ganglia and thalamus is presented.