To test the hypothesis that repeated sessions of anodal or cathodal tDCS in subacute stroke patients produce more improvement in motor function of the affected upper limb than sham 6 months after the intervention. Design: Double-blind, sham-controlled clinical trial. Setting: Neurorehabilitation units at 2 tertiary university hospitals. Participants: Eighteen subacute stroke patients with hand motor impairment. Interventions: Subjects were randomized to receive 10-day sessions of anodal tDCS over the affected motor cortex (2mA for 30 minutes) or cathodal tDCS over the unaffected motor cortex (2mA for 30 minutes), or sham (2mA for 1 minute). Otherwise patients continued their routine treatment. Main Outcome Measures: Blinded raters evaluated motor function using Fugl-Meyer score (FMS) and Modified Barthel Index (MBI) at baseline, immediate after stimulation, 1 month, and 6 months after the intervention. Results: Baseline scores for FMS and MBI were comparable across groups (PϾ.05) However, a repeated measures of analysis of variance revealed a significant intervention ϫ time interaction indicating that anodal and cathodal tDCS had different effects on the FMS and MBI (PϽ.05). Post hoc testing showed that cathodal tDCS led to a greater improvement in FMS than sham at 6 months after the end of intervention (PϽ.05). There was a significant inverse correlation between baseline FMS and FMS gain at 6 months (rϭϪ0.846, Pϭ.001). No patient experienced adverse effects. Conclusions: These results confirm that repeated daily sessions of tDCS over motor cortex result in more improvement of motor function relative to sham, suggesting the potential beneficial role of noninvasive cortical stimulation during rehabilitative motor training.