PurposeThe aim of this study was to determine the comparative accuracy and precision of routine Magnetic Resonance Imaging (MRI) versus MR arthrogram in measuring labral tear size as a function of time from a shoulder dislocation. MethodsWe retrospectively evaluated consecutive patients who underwent primary arthroscopic stabilization between 2012 and 2021 in a single academic center. All patients completed a pre-operative MRI or MR arthrogram of the shoulder within 60 days of injury, and subsequently underwent arthroscopic repair within 6 months of imaging. Intraoperative labral tear size and location was used as standard for comparison. Three musculoskeletal radiologists independently interpreted tear extent using a clock-face convention. Accuracy and precision of MR labral tear measurements were defined based on location and size of the tear, respectively. Accuracy and precision were compared between MRI and MR arthrogram, as a function of time from dislocation. Results32 MRIs and 65 MR arthrograms (total n=97) were assessed. Multivariate analysis demonstrated that intraoperative tear size, early imaging, and arthrogram status were associated with increased MR accuracy and precision (p<0.05). Ordering surgeons preferred arthrogram for delayed imaging (p=0.018). For routine MRI, error in accuracy increased by 3.4° per day and error in precision increased by 2.3° per day (p<0.001) from time of injury. MR arthrogram, however, was not temporally influenced. Significant loss of accuracy and precision of MRI compared to MR arthrogram occurred at 2 weeks after an acute shoulder dislocation. ConclusionCompared to MR Arthrogram, conventional MRI demonstrates time-dependent loss of accuracy and precision in determining shoulder labral tear extent following dislocation, with statistical divergence occurring at 2 weeks.