© 2019 The Authors. Journal of Applied Clinical Medical Physics published by Wiley Periodicals, Inc. on behalf of American Association of Physicists in Medicine Purpose: To evaluate the accuracy of monitoring intrafraction motion during stereotactic radiotherapy with the optical surface monitoring system. Prior studies showing a false increase in the magnitude of translational offsets at non-coplanar couch positions prompted the vendor to implement software changes. This study evaluated two software improvements intended to address false offsets. Methods: The vendor implemented two software improvements: a volumetric (ACO) rather than planar calibration and, approximately 6 months later, an improved calibration workflow (CIB) designed to better compensate for thermal drift. Offsets relative to the reference position, obtained at table angle 0 following image-guided setup, were recorded before beam-on at each table position and at the end of treatment the table returned to 0° for patients receiving SRT. Results: Prior to ACO, between ACO and CIB, and after CIB, 223, 155, and 436 fractions were observed respectively. The median magnitude of translational offsets at the end of treatment was similar for all three intervals: 0.29, 0.33, and 0.27 mm. Prior to ACO, the offset magnitude for non-zero table positions had a median of 0.79 mm and was found to increase with increasing distance from isocenter to the anterior patient surface. After ACO, the median magnitude was 0.74 mm, but the dependence on surface-to-isocenter distance was eliminated. After CIB, the median magnitude for non-zero table positions was reduced to 0.57 mm. Conclusion: Ongoing improvements in software and calibration procedures have decreased reporting of false offsets at non-zero table angles. However, the median magnitude for non-zero table angles is larger than that observed at the end of treatment, indicating that accuracy remains better when the table is not rotated.