Image-guided interventions enable the surgeon to display the position of instruments and devices with respect to the patient's imaging studies during surgery by means of a tracker device. Optical trackers are commonly chosen for many surgical applications when high accuracy and robustness are required. OptiTrack is a multicamera optical tracker whose number of sensors and their spatial configuration can be adapted to the application requirements, making it suitable for surgical settings. Nonetheless, no extensive studies of its accuracy are available. The purpose of this paper was to evaluate an eight-camera optical tracker in terms of accuracy, miscalibration sensitivity, camera occlusions, and tool detection in a feasible clinical setup. We studied the tracking accuracy of the system using a robotic arm (similar to mu m precision) as the gold standard, a single reflective marker, and various tracked objects while the system was installed in an operating room. Miscalibration sensitivity was 0.16 degrees. Mean target error was 0.24 mm for a single marker, decreasing to 0.05 mm for tracked tools. Single-marker error increased up to 1.65 mm when five cameras where occluded although 75% of the working volume showed an error lower than 0.23 mm. The accuracy was sufficient for navigating the collimator in intraoperative electron radiation therapy, improving redundancy and allowing large-working volumes. The tracker assessment we present and the validated miscalibration protocol are important contributions to image-guided surgery, where the choice of the tracker is critical and the knowledge of the accuracy in situations of camera occlusion is mandatory during surgical navigation.