OBJECTIVE Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is widely used in patients with Parkinson's disease (PD). However, which target area of this region results in the highest antiparkinsonian efficacy is still a matter of debate. The aim of this study was to develop a more accurate methodology to locate the electrodes and the contacts used for chronic stimulation (active contacts) in the subthalamic region, and to determine the position at which stimulation conveys the greatest clinical benefit. METHODS The study group comprised 40 patients with PD in whom bilateral DBS electrodes had been implanted in the STN. Based on the Morel atlas, the authors created an adaptable 3D atlas that takes into account individual anatomical variability and divides the STN into functional territories. The locations of the electrodes and active contacts were obtained from an accurate volumetric assessment of the artifact using preoperative and postoperative MR images. Active contacts were positioned in the 3D atlas using stereotactic coordinates and a new volumetric method based on an ellipsoid representation created from all voxels that belong to a set of contacts. The antiparkinsonian benefit of the stimulation was evaluated by the reduction in the Unified Parkinson's Disease Rating Scale Part III (UPDRS-III) score and in the levodopa equivalent daily dose (LEDD) at 6 months. A homogeneous group classification for contact position and the respective clinical improvement was applied using a hierarchical clustering method. RESULTS Subthalamic stimulation induced a significant reduction of 58.0% ± 16.5% in the UPDRS-III score (p < 0.001) and 64.9% ± 21.0% in the LEDD (p < 0.001). The greatest reductions in the total and contralateral UPDRS-III scores (64% and 76%, respectively) and in the LEDD (73%) were obtained when the active contacts were placed approximately 12 mm lateral to the midline, with no influence of the position being observed in the anteroposterior and dorsoventral axes. In contrast, contacts located about 10 mm from the midline only reduced the global and contralateral UPDRS-III scores by 47% and 41%, respectively, and the LEDD by 33%. Using the ellipsoid method of location, active contacts with the highest benefit were positioned in the rostral and most lateral portion of the STN and at the interface between this subthalamic region, the zona incerta, and the thalamic fasciculus. Contacts placed in the most medial regions of the motor STN area provided the lowest clinical efficacy. CONCLUSIONS The authors report an accurate new methodology to assess the position of electrodes and contacts used for chronic subthalamic stimulation. Using this approach, the highest antiparkinsonian benefit is achieved when active contacts are located within the rostral and the most lateral parts of the motor region of the STN and at the interface of this region and adjacent areas (zona incerta and thalamic fasciculus).