Purpose: The purpose of the present study was to evaluate the diagnostic efficacy of CT arthrography (CTA) in the assessment of various shoulder pathologies, compared with MR arthrography (MRA) and MRI with arthroscopic correlation. Materials and Methods: CTA in 84 patients, MRA in 70 patients, and MRI in 27 patients were obtained. A radiologist interpreted each image for 5 pathologies: Bankart, SLAP, Hill-Sachs lesion, full-thickness, and partial-thickness rotator cuff tear. Detailed arthroscopic reports were compared with CTA, MRA, and MRI. The sensitivity, specificity, predictive values, and accuracy were calculated. The agreement between each diagnostic modality and arthroscopy was calculated. Diagnostic efficacy was assessed by the areas under the receiver operating characteristic (ROC) curves. Results: The diagnostic values of all three imaging groups were comparable to each other for Bankart, SLAP, Hills-Sachs, and full-thickness cuff tear lesions, but those of CTA were lower than MRI and MRA for partial-thickness cuff tears. The areas under the ROC curves for CTA, MRA, and MRI were not significantly different for all pathologies, except for partial-thickness cuff tears. Conclusion: CTA was equally competent to MRA or MRI in demonstrating Bankart, Hill-Sachs lesions, SLAP, and full thickness rotator cuff tears but not as efficient in diagnosing partial thickness rotator cuff tears.