It is still unclear why balance impairment in Parkinson's disease (PD) often responds insufficiently to dopaminergic medication. We have studied this issue in 23 patients with idiopathic PD and 24 healthy controls. Our specific purposes were (a) to investigate the contribution of abnormal automatic postural responses to balance impairment in PD and (b) to assess the influence of dopaminergic medication on abnormal automatic postural responses and balance impairment. Standing subjects received 4 degrees "toe-up" rotational perturbations of a supporting forceplate. We bilaterally recorded posturally destabilizing medium latency (ML) responses from the stretched gastrocnemius muscles and functionally corrective long latency (LL) responses from the shortened tibialis anterior (TA) muscles. We also assessed changes in the center of foot pressure (CFP) and the center of gravity (COG). All patients were tested in the "off" and "on" phases. All controls were tested and retested after 1 h. During the off phase, we found enlarged ML amplitudes and diminished LL amplitudes in patients, together with a markedly increased posterior displacement of the COG. The abnormal ML and LL responses were partially responsible for the increased body sway in patients because the initial forward (destabilizing) displacement of the CFP was increased, while the subsequent backward displacement of the CFP (a measure of the corrective braking action of LL responses) was delayed. Abnormal late automatic or possibly more voluntary postural corrections also contributed substantially to the increased body sway. During the on phase, ML amplitudes were reduced in patients but remained increased compared with controls. LL amplitudes no longer differed between both groups due to a modest, possibly dopamine-related increase in patients and a simultaneous decrease in controls. The abnormal CFP displacement was only partially improved by dopaminergic medication. The later postural corrections were not improved at all. Consequently, the increased posterior COG displacement was not ameliorated during the on phase. We conclude that (a) a combination of abnormal automatic and perhaps more voluntary postural corrections contributes to increased body sway in PD and (b) dopaminergic medication fails to improve balance impairment in PD because early automatic postural responses are only partially corrected, while later occurring postural corrections are not improved at all. These electrophysiological results support clinical observations and suggest that nondopaminergic lesions play a significant role in the pathophysiology of postural abnormalities in PD.