Despite optimum medical management, many patients with Parkinson's disease are incapacitated by gait disorders including freezing of gait. We aimed to assess whether methylphenidate--through its combined action on dopamine and noradrenaline reuptake--would improve gait disorders and freezing of gate in patients with advanced Parkinson's disease without dementia who also received subthalamic nucleus stimulation. This multicentre, parallel, double-blind, placebo-controlled, randomised trial was done in 13 movement disorders departments in France between October, 2009, and December, 2011. Eligible patients were younger than 80 years and had Parkinson's disease, severe gait disorders, and freezing of gate despite optimised treatment of motor fluctuations with dopaminergic drugs and subthalamic stimulation. We randomly assigned patients (1:1 with a computer random-number generator in blocks of four) to receive methylphenidate (1 mg/kg per day) or placebo capsules for 90 days. Patients, their carers, study staff, investigators, and data analysts were masked to treatment allocation. To control for confounding effects of levodopa we assessed patients under standardised conditions with an acute levodopa challenge. Our primary outcome was a change in the number of steps during the stand-walk-sit (SWS) test without levodopa. We compared the respective mean numbers of steps at day 90 in the methylphenidate and placebo groups in a covariance analysis and adjusted for baseline differences. This trial is registered with ClinicalTrials.gov, number NCT00914095. We screened 81 patients and randomly assigned 35 to receive methylphenidate and 34 to receive placebo. 33 patients in the methylphenidate group and 32 patients in the placebo group completed the study. Efficacy outcomes were assessed in the patients who completed the study. Compared with patients in the placebo group (median 33 steps [IQR 26-45]), the patients in the methylphenidate group made fewer steps at 90 days (31 [26-42], F((1, 62))=6·1, p=0·017, adjusted size effect 0·61). Adverse events were analysed in all randomly assigned patients. There were significantly more adverse events in the methylphenidate group compared with placebo. Patients on methylphenidate had a significant increase in heart rate (mean 3·6 [SD 7·2] beats per min) and decrease in weight (mean 2·2 [SD 1·8] kg) compared with the placebo group. Methylphenidate improved gait hypokinesia and freezing in patients with advanced Parkinson's disease receiving subthalamic nucleus stimulation. Methylphenidate represents a therapeutic option in the treatment of gait disorders at the advanced stage of Parkinson's disease. The long term risk-benefit balance should be further studied. French Ministry of Health and Novartis Pharma.