17005 Background: The on-board megavoltage (MV) computed tomography (CT) capabilities of a Tomotherapy unit were used to obtain the daily MVCT images of lung cancer patients. For daily patient alignment, differences between the MVCT scan and planning CT were resolved by calculating the necessary couch shifts in the X = mediolateral, Y = craniocaudal, and Z = anteriorposterior directions. Daily shifts were analyzed. Methods: 583 alignments from 36 patients with lung cancer were available for analysis. The systematic (Σ) and random (σ) errors were calculated and a covariate analysis was performed with tumor size, Karnofsky Performance Score (KPS), and presence of atelectasis. Two error minimization strategies were applied to the data - 1) shifts from fraction 1 were subtracted from subsequent shifts, and 2) the average of shifts 1–3 were subtracted from shifts 4 onward. Σ and σ were calculated for each of the 3 data sets and applied to van Herk’s margin recipe 2.5 Σ + 0.7σ. The mean, standard deviation, and standard error of the magnitude shifts for 13 patients who each received 23 fractions were analyzed by Spearman’s rank correlation test for the relationship between shift magnitude and fraction number. Results: The presence of atelectasis was significantly related to a smaller σ in millimeters, 2.8 ± 0.08 vs. 3.5 ± 0.09 (p = 1.1 × 10−8). The other covariates were not significantly related to set-up error. The 2nd error minimization strategy decreased Σ in the X, Y, and Z directions from 4.7 ± 0.6, 5.8 ± 0.9, 4.9 ± 0.6 to 2.1 ± 0.1, 4.2 ± 0.5, 3.4 ± 0.3 (p = 2.0 × 10−5, 0.13, 0.02) respectively. Calculated margins from van Herk’s equation for all data reported as (x, y, z) in mm were (13.8, 19.6, and 15.9). For strategies 1 and 2 respectively, calculated margins were reduced by (27.2%, 11.5%, 10.6%) and (46.7%, 21.5%, 23.2%). The mean magnitude of isocenter shift and the standard deviation were found to increase with fraction number (p = 1.0 × 10−6 and 5.0 × 10−5 respectively). Conclusion: The error correction strategies significantly reduced Σ but did not reduce the margins dramatically. Drift in accuracy during a long treatment course and an inability to identify subgroups of patients based on our covariates who may not need daily imaging suggests that daily image verification + correction will help reduce error and margins. No significant financial relationships to disclose.