A typical pathophysiological abnormality in dystonia is cocontraction of antagonist muscles, with impaired reciprocal inhibitory mechanisms in the spinal cord. Recent experimental data have shown that inhibitory interactions between antagonist muscles have also a parallel control at the level of the sensorimotor cortex. The aim of this work was to study heteronymous effects of a median nerve stimulus on the corticospinal projections to forearm muscles in dystonia. We used the technique of antagonist cortical inhibition, which assesses the conditioning effect of median nerve afferent input on motor evoked potentials (MEPs) elicited by transcranial magnetic stimulation (TMS) in ipsilateral forearm extensor muscles at rest. Nine healthy subjects and 10 patients with torsion dystonia participated in the study. MEPs and somatosensory evoked potentials were normal in patients. In healthy subjects, median nerve stimulation at 15- to 18-msec intervals inhibited the test MEPs in forearm extensors. In dystonic patients, median nerve stimulation delivered at the same conditioning-test intervals elicited significantly less inhibition of the test MEP. On the whole, these data suggest an impaired sensory-motor integration in dystonia and, more specifically, the decreased antagonistic cortical inhibition could suggest that functional interactions between antagonist muscles are primarily impaired at the cortical level.