The CyberKnife system generates log-files including actual treatment parameters for each procedure. In this study, log-files were analyzed to evaluate the mechanical uncertainty in beam localization over the long term (approximately 1 year), as were patterns of patient movements during brain tumor treatments using CyberKnife.The clinical to planning target volume (CTV-PTV) margin in clinical use was examined based on this analysis. Log-file analysis was performed using data from 140 brain tumor patients (267 treatment plans; 27,166 beams; approximately 66 beams/fraction), who underwent CyberKnife stereotactic radiosurgery and radiation therapy. We calculated a mean error and 2 standard deviations (2σ) for this population. Additionally, we calculated the radius R95% spatially covering 95% of all error vectors.The mean mechanical uncertainties of CyberKnife brain tumor treatment were found to be 0.07, 0.01, and -0.09 mm in the +inferior/-superior, +left/-right, and +anterior/-posterior directions, respectively. The mean (2σ) of R95% was 1.02 (0.42) mm. A smaller degree of correlation between patient movement and R95% was observed.The CyberKnife is robust in tracking accuracy, regardless of patient movement. The effectiveness of log-file analysis was demonstrated regarding quality control for monitoring beam localization in the CyberKnife system. The CTV-PTV margin of 2.0 mm was found to be adequate in brain tumor treatments using the CyberKnife.