OBJECTIVE: The optimal imaging modality for preoperative targeting of the subthalamic nucleus (STN) for high-frequency stimulation is controversially discussed. Commonly used methods were stereotactic magnetic resonance imaging (MRI), stereotactic ventriculography, and fusion between MRI and stereotactic computer tomography. All of these techniques not only have their own advantages but also specific limitations and drawbacks. The purpose of this study was to evaluate the accuracy of the preoperative MRI targeting as compared with ventriculography in terms of both the STN target as well as the internal landmarks. METHODS: Thirty patients with Parkinson's disease who underwent bilateral surgery for STN-high-frequency stimulation received both stereotactic ventriculography and stereotactic MRI. The theoretical target was determined by each of these two imaging modalities. The final electrode placement was performed after extensive electrophysiological evaluation using microrecording and microstimulation. The real target was assumed to be given by the electrode contact with the best clinical result assessed by the United Parkinson's Disease Rating Scale in the postoperative follow-up. In addition, the coordinates of the two landmarks, anterior commissure and posterior commissure, were determined using both imaging methods. RESULTS: The mean targeting error was 4.1 +/- 1.7 mm (mean +/- standard deviation) for MRI and 2.4 +/- 1.1 mm for ventriculography (P < 0.0001). The mean target mismatch between the two imaging methods was 2.9 +/- 1.2 mm. The length of the anterior commissure-posterior commissure distance differed significantly (P < 0.0001) between MRI (27.6 +/- 1.6 mm) and ventriculography (25.0 +/- 1.3 mm). The mismatch was mainly induced by an anterior diplacement of the anterior commissure by 1.9 +/- 2.2 mm (P < 0.0001) in MRI determination, as compared with ventriculography. CONCLUSION: Our findings show that the indirect targeting of the STN using coordinates based on radiological landmarks is more accurate than the direct targeting using anatomic visualization of the target structure. Regardless of the imaging procedure, electrophysiological mapping is required for optimal electrode placement, although in 20% of cases, the target determined by MRI falls out of the radius explored by electrophysiology.