This multicentre randomised double-blind 3- to 5-year trial was designed to assess whether initial therapy with cabergoline alone or in combination with levodopa prevents or delays the occurrence of long term motor complications in patients with early Parkinson's disease. Patients eligible for study inclusion (n = 412) had early idiopathic Parkinson's disease (Hoehn and Yahr stages 1 to 3) and had received no previous treatment with levodopa, selegiline or dopamine agonists. Patients were randomised to receive either cabergoline (0.25 to 4 mg once daily) or levodopa (100 to 600 mg/day) titrated over a maximum period of 24 weeks. Once the optimum or maximum tolerated dose was achieved, it was maintained up to the end-point (development of motor complications confirmed at 2 consecutive 3-month visits) or up to a minimum of 3 years' treatment. Open labelled levodopa was added in both treatment arms when the improvement in motor disability [Unified Parkinson's Disease Rating Scale (UPDRS) factor III] decreased below 30% vs baseline. Both treatments improved motor disability, decreasing UPDRS factor III scores and factor II scores for activities of daily living. The development of motor complications (end-point) was significantly less frequent in patients treated with cabergoline than in levodopa recipients (22% vs 34%; p < 0.02). The relative risk of developing motor complications during treatment with cabergoline was more than 50% lower than with levodopa. Serious adverse events, either drug related or not, were slightly more frequent in cabergoline-treated patients (31%) than in those treated with levodopa (25%). The withdrawal rate in the cabergoline vs levodopa group was 16 vs 13%. In conclusion, the study shows that, in patients with early Parkinson's disease, cabergoline is effective either as monotherapy or combined with levodopa. Moreover, starting treatment with cabergoline significantly delays the development of motor complications.