The long term effects of a first line treatment with levodopa or bromocriptine were compared in 36 previously untreated patients with Parkinson's disease in a prospective randomized trial: 18 patients were treated with levodopa alone (mean dose: 485 +/- 71 mg daily) whereas 18 others received bromocriptine alone (mean dose: 55 +/- 6 mg daily) during 36 +/- 3 and 31 +/- 3 months respectively. We observed a similar decrease in the Columbia rating scale but the nature of long term side effects was different in the two groups: patients on levodopa developed peak-dose dyskinesias (5 cases), wearing off akinesia (1 case) and on-off effects (1 case). Under bromocriptine treatment, 2 patients developed severe psychosis whereas one suffered from primary lack of drug effectiveness and 5 others from late decrease of drug effectiveness. These results suggest the potential value of relatively high doses of D2 dopamine agonists (such as bromocriptine) in the first line treatment of Parkinson's disease: however, their use can be limited by their decrease of effectiveness after several years and/or the occurrence of severe neuropsychiatric side-effects.