Previous short-term studies have shown that the dopamine agonist pergolide improves control of Parkinson's disease when used in conjunction with carbidopa-levodopa (Sinemet). We assessed the long-term outcome (2 1/2- to 3-year follow-up) in patients with Parkinson's disease who participated in our previous pergolide double-blind trial and were subsequently switched to open-label pergolide therapy. Of 41 evaluable patients who began pergolide therapy, 10 (24%) experienced sustained substantial benefit that persisted to the end of this investigation. A total of 23 patients (56%) remained on pergolide therapy and, as a group, had considerable improvement over baseline at 2 1/2 to 3 years on the basis of several measurements of efficacy. A tendency toward deterioration could be reversed in many patients by larger or more frequent doses of carbidopa-levodopa; nevertheless, all but four patients were still taking the same dose or less of carbidopa-levodopa at the end of this study as at the onset. Patients with the best initial response to pergolide seemed most likely to experience long-term benefit. Confusion and hallucinations were the side effects most likely to necessitate discontinuation of pergolide. Symptoms suggestive of dose-related angina pectoris occurred in four patients in the open-label phase and two patients in the earlier double-blind phase (13% of patients who started pergolide therapy); these symptoms were easily controlled by dose reduction or discontinuation of pergolide, without sequelae. Dose-related leukopenia developed in one patient.