Background: To assess the accuracy and precision of a fully integrated pilot installation of stereoscopic X-ray imaging and kV-CBCT for automatic couch positioning in stereotactic radiotherapy of intracranial tumors. Positioning errors as detected by stereoscopic X-ray imaging are compared to those by kV-CBCT (i.e. the accuracy of the new method is verified by the established method), and repeated X-ray images are compared (i.e. the precision of new method is determined intra-modally). Methods: Preliminary results are reported from a study with 32 patients with intracranial tumors. Patients were treated with stereotactic radiotherapy guided by stereoscopic X-ray imaging and kV-CBCT. Patient positioning was automatically corrected by a robotic couch. Cross-modal discrepancies in position detection were measured (N = 42). Intra-modal improvements after correction and re-verification by stereoscopic X-ray imaging were measured (N = 70). The accuracy and precision of stereoscopic X-ray imaging and the accuracy and precision of CBCT were confirmed in phantom measurements (N = 12 shifts of a ball bearing phantom, N = 24 shifts of a head phantom). Results: After correction based on stereoscopic X-ray imaging 95% of residual mean errors were below 0.4, 0.4, 0.5, and 0.7 mm (lateral, longitudinal, vertical, radial, respectively). Stereoscopic X-ray imaging and CBCT were in close agreement with an average discrepancy of 0.1, 0.5, 0.3 and 0.8 mm, respectively. 95% of discrepancies were below 0.8, 1.2, 1.0, and 1.4 mm, respectively. After correction and re-verification by stereoscopic X-ray imaging, the remaining intra-modal residual error was consistent with zero (p = 0.31, p = 0.48, p = 0.81 in lateral, longitudinal, and vertical direction;p-values from two-tailed t-test). The inherent technical accuracy and precision of stereoscopic X-ray imaging and the accuracy and precision of CBCT were found to be of the order of 0.1 mm in controlled phantom settings. Conclusions: In a routine clinical setting, both stereoscopic X-ray imaging and CBCT were able to reduce positioning errors by an order of magnitude. The end-to-end precision of the system, measured from the discrepancy (mean) between ExacTrac and CBCT, in a clinical setting seems to be about 0.8 mm radially, including couch positioning. The precision (measured from repeatability of ExacTrac, intra-modal) was found to be about 0.7 mm radially in a clinical setting.