[en] Purpose: Frameless Stereotactic radiosurgery (SRS) is increasingly used in the clinic. Cone-Beam CT (CBCT) to simulation-CT match has replaced the 3-dimensional coordinate based set up using a stereotactic localizing frame. The SRS frame however served as both a localizing and immobilizing device. We seek to measure the quality of frameless (mask based) and frame based immobilization and evaluate its impact on target dose. Methods: Each SRS patient was set up by kV on-board imaging (OBI) and then fine-tuned with CBCT. A second CBCT was done at treatment-end to ascertain intrafraction motion. We compared pre- vs post-treatment CBCT shifts for both frameless and frame based SRS patients. CBCT to sim-CT fusion was repeated for each patient off-line to assess systematic residual image registration error. Each patient was re-planned with measured shifts to assess effects on target dose. Results: We analyzed 11 patients (12 lesions) treated with frameless SRS and 6 patients (11 lesions) with a fixed frame system. Average intra-fraction iso-center positioning errors for frameless and frame-based treatments were 1.24 ± 0.57 mm and 0.28 ± 0.08 mm (mean ± s.d.) respectively. Residual error in CBCT registration was 0.24 mm. The frameless positioning uncertainties led to target dose errors in Dmin and D95 of 15.5 ± 18.4% and 6.6 ± 9.1% respectively. The corresponding errors in fixed frame SRS were much lower with Dmin and D95 reduced by 4.2 ± 6.5% and D95 2.5 ± 3.8% respectively. Conclusion: Frameless mask provides good immobilization with average patient motion of 1.2 mm during treatment. This exceeds MRI voxel dimensions (∼0.43mm) used for target delineation. Frame-based SRS provides superior patient immobilization with measureable movement no greater than the background noise of the CBCT registration. Small lesions requiring submm precision are better served with a frame based SRS.