BACKGROUND: Depression is common in patients with Parkinson's disease, but evidence on the efficacy of antidepressants in this population is lacking. Because depression in patients with Parkinson's disease might be related to dopaminergic dysfunction, we aimed to assess the efficacy of the dopamine agonist pramipexole for treatment of depressive symptoms in patients with Parkinson's disease. METHODS: We did a 12-week randomised, double-blind, placebo-controlled (1:1 ratio) trial of pramipexole (0.125-1.0 mg three times per day) compared with placebo in patients with mild-to-moderate Parkinson's disease. Patients from 76 centres in 12 European countries and South Africa were included if they were on stable antiparkinsonian therapy without motor fluctuations and had depressive symptoms (15-item geriatric depression scale score > or =5 and unified Parkinson's disease rating scale [UPDRS] part 1 depression item score > or =2). Patients were randomly assigned by centre in blocks of four by use of a randomisation number generating system. Clinical monitors, the principal investigator, and patients were masked to treatment allocation. The primary endpoint was change in Beck depression inventory (BDI) score and all treated patients who had at least one post-baseline efficacy assessment were included in the primary analysis. We also did a pre-specified path analysis with regression models to assess the relation between BDI and UPDRS part 3 (motor score) changes. This trial is registered with ClinicalTrials.gov, number NCT00297778, and EudraCT, number 2005-003788-22. FINDINGS: Between March, 2006, and February, 2008, we enrolled 323 patients. Of 296 patients randomly assigned to pramipexole or placebo, 287 were included in the primary analysis: 139 in the pramipexole group and 148 in the placebo group. BDI scores decreased by an adjusted mean 5.9 (SE 0.5) points in the pramipexole group and 4.0 (0.5) points in the placebo group (difference 1.9, 95% CI 0.5-3.4; p=0.01, ANCOVA). The UPDRS motor score decreased by an adjusted mean 4.4 (0.6) points in the pramipexole group and 2.2 (0.5) points in the placebo group (difference 2.2, 95% CI 0.7-3.7; p=0.003, ANCOVA). Path analysis showed the direct effect of pramipexole on depressive symptoms accounted for 80% of total treatment effect (p=0.04). Adverse events were reported in 105 of 144 patients in the pramipexole group and 101 of 152 in the placebo group. Adverse events in the pramipexole group were consistent with the known safety profile of the drug. INTERPRETATION: Pramipexole improved depressive symptoms in patients with Parkinson's disease, mainly through a direct antidepressant effect. This effect should be considered in the clinical management of patients with Parkinson's disease.