Deep brain stimulation of different targets has been shown to drastically improve symptoms of a variety of neurological conditions. However, the occurrence of disabling side effects may limit the ability to deliver adequate amounts of current necessary to reach the maximal benefit. Computed models have suggested that reduction in electrode size and the ability to provide directional stimulation could increase the efficacy of such therapies. This has never been demonstrated in humans. In the present study, we assess the effect of directional stimulation compared to omnidirectional stimulation. Three different directions of stimulation as well as omnidirectional stimulation were tested intraoperatively in the subthalamic nucleus of 11 patients with Parkinson's disease and in the nucleus ventralis intermedius of two other subjects with essential tremor. At the trajectory chosen for implantation of the definitive electrode, we assessed the current threshold window between positive and side effects, defined as the therapeutic window. A computed finite element model was used to compare the volume of tissue activated when one directional electrode was stimulated, or in case of omnidirectional stimulation. All but one patient showed a benefit of directional stimulation compared to omnidirectional. A best direction of stimulation was observed in all the patients. The therapeutic window in the best direction was wider than the second best direction (P = 0.003) and wider than the third best direction (P = 0.002). Compared to omnidirectional direction, the therapeutic window in the best direction was 41.3% wider (P = 0.037). The current threshold producing meaningful therapeutic effect in the best direction was 0.67 mA (0.3-1.0 mA) and was 43% lower than in omnidirectional stimulation (P = 0.002). No complication as a result of insertion of the directional electrode or during testing was encountered. The computed model revealed a volume of tissue activated of 10.5 mm(3) in omnidirectional mode, compared with 4.2 mm(3) when only one electrode was used. Directional deep brain stimulation with a reduced electrode size applied intraoperatively in the subthalamic nucleus as well as in the nucleus ventralis intermedius of the thalamus significantly widened the therapeutic window and lowered the current needed for beneficial effects, compared to omnidirectional stimulation. The observed side effects related to direction of stimulation were consistent with the anatomical location of surrounding structures. This new approach opens the door to an improved deep brain stimulation therapy. Chronic implantation is further needed to confirm these findings.