Although frameless stereotactic techniques have become indispensable in neurosurgery, their technical complexity requires careful definition and evaluation. Navigation is of particular concern when it is applied to approach a complex, tight surgical area like the temporal bone, where every millimeter is important. Theoretically, the temporal bone is an ideal location in which to use image-guided navigation because its bony construct precludes pre- and intraoperative shift. In this context, the feasibility of using a navigational system is determined by the system's accuracy and by the spatial characteristics of the targets. Literature addressing the accuracy of image guidance techniques in temporal bone surgery is relatively sparse. Accuracy of these systems within the temporal bone is still under investigation. We investigated the application accuracy of computed tomography-based, frameless, image-guided navigation to identify various bony structures in the temporal bone via a retrosigmoid approach. In a total of 10 operations, we performed a retrosigmoid approach simulating operative conditions on either side of 5 whole, fresh cadaveric heads. Six titanium microscrews were implanted around the planned craniotomy site as permanent bone reference markers before the surgical procedure. High-resolution computed tomographic scans were obtained (slice thickness, 0.6-mm, contiguous non-overlapping slices; gantry setting, 0 degrees; scan window diameter, 225 mm; pixel size, >0.44 x 0.44). We used a Stryker navigation system (Stryker Instruments, Kalamazoo, MI) for intraoperative navigation. External and internal targets were selected for calculation of navigation accuracy. The system calculated target registration error to be 0.48 +/- 0.21 mm, and the global accuracies (navigation accuracies) were calculated using external over-the-skull and internal targets within the temporal bone. Overall navigation accuracy was 0.91 +/- 0.28 mm; for reaching internal targets within temporal bone, accuracy was 0.94 +/- 0.22 mm; and for external targets, accuracy was 0.83 +/- 0.11 mm. Ninety-five percent of targets could be reached within 1.4 mm of their actual position. Using high-resolution computed tomography and bone-implanted reference markers, frameless navigation can be as accurate as frame-based stereotaxy in providing a guide to maximize safe surgical approaches to the temporal bone. Although error-free navigation is not possible with the submillimetric accuracy required by direct anatomic contouring of tiny structures within temporal bone, it still provides a road map to maximize safe surgical exposure.