Vibratory inhibition, the homonymous recovery curve and the ratio of the maximal H-reflex to direct muscle potential (H/M ratio) of the soleus H-reflex were assessed in 10 patients with leg dystonia and in six patients with arm or neck dystonia. The results were compared with those obtained in 48 healthy control subjects. H-reflex variables most helpful for the discrimination of patients and healthy subjects were identified. In patients with leg dystonia, vibratory inhibition was less marked than in control subjects, whereas late facilitation of the recovery curve was increased. In patients with leg dystonia, area values of test reflexes in the late facilitatory phase of the recovery curve exceeded peak-peak values, in contrast to findings in control subjects. This finding may be attributable to less synchronization of enhanced test reflexes in dystonia than in the control condition. In differentiating patients with leg dystonia from control subjects, a combination of parameters of vibratory inhibition and the late facilitatory phase of the recovery curve appeared most useful. In patients with arm or neck dystonia and in the unaffected legs of hemidystonic patients, soleus H-reflex test results were in the normal range. Abnormalities in the results of the soleus H-reflex tests we used appear to be related to the presence of clinical signs in the extremity under examination and not to the severity of features.