Somatosensory evoked potentials (SEPs) of the median nerve were recorded in 40 patients suffering from cervical dystonia and in 40 healthy volunteers as a control. Before recording the median nerve SEPs, polymyographic recordings were performed in all patients with cervical dystonia. The activity of cervical muscles was recorded, and the leading muscle of cervical dystonia was determined. Patients were divided into two groups according to the results of polymyography. The leading muscle was sternocleidomastoid in the first group and the splenius capitis in the second group. Patient SEPs were recorded during abnormal head rotation. SEPs in 20 healthy volunteers were recorded with the head in the middle position. SEPs of another 20 healthy volunteers were recorded with the head rotated 60 degrees to the right. The mean peak-to-peak amplitude values of the precentral P22/N30 complex and the mean ratio of the P22/N30 amplitudes between both hemispheres were calculated in the F3 (F4) and C3' (C4') electrode positions in all four groups. No significant lateralization of the precentral P22/ N30 component was found in either group of healthy volunteers. In dystonic patients in whom the sternocleidomastoid was determined as the leading muscle of dystonia, a statistically significant lateralization of the P22/N30 component toward the ipsilateral side of the leading muscle was found. In the group with the splenius capitis determined as the leading muscle of dystonia, a statistically significant lateralization of the P22/N30 component toward the contralateral side of the leading muscle was found. The possibility that the precentral cortex is activated differently in cervical dystonia patients who have different muscle patterns of dystonia is discussed.