We made a double-blind, crossover comparison of buspirone (10 mg orally, twice a day) and placebo in 10 patients with Parkinson's disease and levodopa-induced dyskinesias. The duration of the study was 3 weeks, for both buspirone and for placebo treatment. Chronic therapies remained unchanged. The extrapyramidal symptoms, dyskinesias, and anxious/depressive symptoms were evaluated at the beginning of the study and after the buspirone and placebo treatments. Seven patients concluded the trial. The extrapyramidal symptoms, evaluated in both the "off" and "on" states during an oral L-Dopa test, did not show any worsening during the trial. Buspirone significantly lessened the severity of levodopa-induced dyskinesias in five of the seven patients, whereas it proved ineffective in the two patients with the mildest dyskinesias. There were no changes in the mild anxious and depressive symptoms at any time during the study. The activities of buspirone on the serotonin and dopamine systems might have led to the antidyskinetic effect we observed. At daily dosages of 20 mg, buspirone might prove effective in reducing levodopa-induced dyskinesias without worsening of parkinsonism.