High frequency stimulation of the ventral intermedius nucleus (Vim) of the thalamus is successfully used for the treatment of postural tremor. Target coordinates are most commonly calculated using a statistical method. Here, we compare a statistical and an individual targeting method, using an histology-based three-dimensional deformable brain atlas which allows localization of the Vim on individual patient's MR images by adaptation of the atlas onto the patient's brain. Twenty-nine consecutive patients had electrodes implanted in the Vim uni-or bilaterally for severe essential tremor. Thirty-five targets were determined by calculating the statistical target and then using the deformable atlas to compute the individual target. Pythagorean distance between these targets was calculated. Statistical and individual targets were compared by double blind evaluation of perioperative stimulation effects. For most cases (n = 24), the Pythagorean distance was higher than 1.5 mm. In 79% of these cases, the definitive electrode was implanted using the position of the individual target. For the remaining cases (n = 11, distance < 1.5 mm), the definitive electrode was implanted according to the statistical target location in 73% of the cases. As a whole, when individual target was used, it was located at least 2 mm more medial than the statistical one in 86% cases. These results suggest that Vim target determination based on a statistical method might be inaccurate. In particular, laterality might be overestimated, leading to nonoptimal clinical results. In clinical practice, this means that microelectrode exploration during Vim surgery should include at least one trajectory more medial than the statistical target.