Carbamazepine or placebo were given in random order, during two four-week periods, in a double-blind, cross-over trial in six patients presenting with symptoms of restless legs. On global evaluation after completing the trial three patients and the physician preferred to continue treatment with carbamazepine. In the remaining three cases both the physician and the patients preferred not to continue with either of the treatments. However, the patients who did not give any preference also had less pronounced symptoms during treatment with carbamazepine. No patient felt better during treatment with placebo as compared to carbamazepine. The results indicate that certain patients have fewer attacks of restless legs during treatment with carbamazepine.