OBJECTIVE The actual distortion present in a given series of magnetic resonance (MR) images is difficult to establish. The purpose of this study was to validate an MR imaging-based methodology for stereotactic targeting of the internal globus pallidus during electrode implantation in children in whom general anesthesia had been induced. METHODS Twelve children (mean follow up 1 year) suffering from generalized dystonia were treated with deep brain stimulation by using a head frame and MR imaging. To analyze the influence of distortions at every step of the procedure, the geometrical characteristics of the frame were first controlled using the localizer as a phantom. Then pre- and postoperative coordinates of fixed anatomical landmarks and electrode positions, both determined with the head frame in place, were statistically compared. No significant difference was observed between theoretical and measured dimensions of the localizer (Student's t-test, ¿t¿ > 2.2 for 12 patients) in the x, y, and z directions. No significant differences were observed (Wilcoxon paired-sample test) between the following: 1) pre- and postoperative coordinates of the anterior commissure (AC) (deltax = 0.3+/-0.29 mm and deltay = 0.34+/-0.32 mm) and posterior commissure (PC) (deltax = 0.15+/-0.18 mm and deltay = 0.34+/-0.25 mm); 2) pre- and postoperative AC-PC distance (deltaL = 0.33+/-0.22 mm); and 3) preoperative target and final electrode position coordinates (deltax = 0.24+/-0.22 mm; deltay = 0.19+/-0.16 mm). CONCLUSIONS In the authors' center, MR imaging distortions did not induce detectable errors during stereotactic surgery in dystonic children. Target localization and electrode implantation could be achieved using MR imaging alone after induction of general anesthesia. The remarkable postoperative improvement in these patients confirmed the accuracy of the procedure (Burke-Marsden-Fahn Dystonia Rating Scale score delta = -83.8%).