Background: The effects of constant-current deep brain stimulation (DBS) have not been studied in controlled trials in patients with Parkinson's disease. We aimed to assess the safety and efficacy of bilateral constant-current DBS of the subthalamic nucleus. Methods: This prospective, randomised, multicentre controlled trial was done between Sept 26, 2005, and Aug 13, 2010, at 15 clinical sites specialising in movement disorders in the USA. Patients were eligible if they were aged 18-80 years, had Parkinson's disease for 5 years or more, and had either 6 h or more daily off time reported in a patient diary of moderate to severe dyskinesia during waking hours. The patients received bilateral implantation in the subthalamic nucleus of a constant-current DBS device. After implantation, computer-generated randomisation was done with a block size of four, and patients were randomly assigned to the stimulation or control group (stimulation:control ratio 3:1). The control group received implantation without activation for 3 months. No blinding occurred during this study, and both patients and investigators were aware of the treatment group. The primary outcome variable was the change in on time without bothersome dyskinesia (ie, good quality on time) at 3 months as recorded in patients' diaries. Patients were followed up for 1 year. This trial is registered with ClinicalTrials.gov, number NCT00552474. Findings: Of 168 patients assessed for eligibility, 136 had implantation of the constant-current device and were randomly assigned to receive immediate (101 patients) or delayed (35 patients) stimulation. Both study groups reported a mean increase of good quality on time after 3 months, and the increase was greater in the stimulation group (4·27 h vs 1·77 h, difference 2·51 [95% CI 0·87-4·16]; p=0·003). Unified Parkinson's disease rating scale motor scores in the off-medication, on-stimulation condition improved by 39% from baseline (24·8 vs 40·8). Some serious adverse events occurred after DBS implantation, including infections in five (4%) of 136 patients and intracranial haemorrhage in four (3%) patients. Stimulation of the subthalamic nucleus was associated with dysarthria, fatigue, paraesthesias, and oedema, whereas gait problems, disequilibrium, dyskinesia, and falls were reported in both groups. Interpretation: Constant-current DBS of the subthalamic nucleus produced significant improvements in good quality on time when compared with a control group without stimulation. Future trials should compare the effects of constant-current DBS with those of voltage-controlled stimulation. Funding: St Jude Medical Neuromodulation Division.