Purpose: RapidArc is a more complicated technique than fixed field IMRT both in terms of treatment planning and delivery. Due to its speedy delivery, RapidArc (RA) is becoming popular in the clinics. Previous work has shown that the detectors used to perform patient specific IMRT QA cannot detect MLC leaf position error induced dosimetric differences even large as 5.6%. We present a patient specific QA technique with the ability to detect MLC leaf error induced dosimetric errors as small as 0.5%. Methods: Dual arc RA plans were made for treatment of head and neck, gastrointestinal and spine cancers of 10 patients on a Trilogy unit. For each patient, four plans were created: i) clinically acceptable RA plan ii) RA plan with a single MLC leaf shifted for only three consecutive control points iii) delivered original plan recreated using linac log file measurements iv) delivered shifted plan recreated using linac log file measurements. Delivered plans were recreated using only MLC positional errors from linac log file measurements. The results were analyzed by comparing voxel by voxel 3D dose differences. To quantify the actual dose difference errors due to intentional MLC shifts, a comparison was performed between the shifted dose map and the original dose map. Results: Simulated MLC shifts gave rise to a trajectory of dose differences with maximum percentage dose differences mean (range) of 0.52 (0.31 – 0.80)%, while measurements from linac log files show a maximum percentage dose differences mean (range) of 0.56 (0.31 – 0.81)%. Maximum background percentage dose differences mean (range) were 0.19 (0.05 – 0.35)%. Measurement related 3D dose map had an average signal to noise ratio of 3.89 for the set of patients analyzed. Conclusion: Linac log file measurements provide the means to detect MLC shift induced dosimetric deviations as small as 0.5% from the prescription dose. UVA Radiation Oncology George Amarino Grants ‐ 2011