Two modes of TDCS have been used in human stroke rehabilitation studies, namely, anodal stimulation (increase in excitability) of the lesional hemisphere (Figure 2) and cathodal stimulation (decrease in excitability) of the contralesional hemisphere. Proof-of-principle studies have been performed for both of these approaches using TMS and TDCS. These studies mostly applied a single session of TMS or TDCS and evaluated the effects, comparing performance in preintervention and postintervention batteries of motor assessments. Effects of multiple sessions are being studied. Preliminary findings of an ongoing trial at our institution involving 5 days of combined TDCS with occupational therapy in a crossover sham-control study26suggested significant improvement in motor outcomes that lasted for at least 1 week. However, results of this cathodal TDCS study (stimulation applied to the contralesional hemisphere) contrast with those of an anodal TDCS study by Hesse et al,27who subjected patients after subacute stroke to multiple sessions of anodal TDCS (applied to the lesional hemisphere) in combination with a robot-assisted arm training protocol but failed to find significant motor improvements. These differences between cathodal stimulation to the unaffected hemisphere and anodal stimulation to the lesional hemisphere may be due to factors such as extent of the lesion, amount of cortical involvement, or involvement of the pyramidal tract on the lesional hemisphere. Further studies, and possibly direct contrasts between cathodal and anodal stimulation approaches, are needed to explore these issues. Previous findings in patients with chronic stroke using behavioral variables and TMS as a diagnostic tool have shown that anodal TDCS applied to the lesional motor region is associated with significant improvements in motor tasks, and the improvements correlated with the increase in excitability of the lesional hemisphere as indicated by a rise in the slope of the recruitment curve and a reduction in the short-interval intracortical inhibition as evidenced by TMS.28Similar findings have recently been made in our group by applying cathodal stimulation to the contralesional unaffected hemisphere in patients with chronic stroke; improvements in motor tasks correlated with a rise in the slope of the recruitment curve in the affected hemisphere and a decrease in the activation of the contralesional hemisphere as revealed by analysis of functional magnetic resonance imaging data.26Future studies might be able to use pretherapy assessments (eg, lesion size and location, integrity of the pyramidal tract, and the presence of abnormal interhemispheric inhibition) to tailor stimulation variables to patients after stroke. Such variables include mode of the stimulation (eg, anodal vs cathodal), strength of the stimulation, region of the brain to which stimulation should be delivered, and the extent of this region that is being stimulated. Transcranial direct current stimulation of the unaffected hemisphere may have the following inherent advantages over stimulation of the affected hemisphere: normal topography, intact intracortical connections, less risk of triggering a seizure (“scar epilepsy”), and reliance on a model of distribution in current density that is not disturbed by a lesion. Apart from the site of stimulation and the lesion size and location, many other factors can contribute to variability in natural and facilitated stroke recovery studies. Among others, these include age, sex, severity of the initial impairment, hemisphere affected (right vs left and dominant vs nondominant), lesion site (eg, cortical or subcortical vs deep white matter lesions), and relation between lesion location and retained pyramidal tract. The integrity of the pyramidal tract as examined using diffusion tensor imaging or as indicated by the presence of motor-evoked potentials in the affected hand is an important determinant of recovery and a predictor of stroke recovery potential.