OBJECTIVE:To evaluate the diagnostic yield of two acquisitions of single-contrast CT arthrography (CTA) of the shoulder in internal, neutral, or external glenohumeral rotation with arthroscopic correlation. MATERIALS AND METHODS:The CT study was obtained using two acquisitions (first the humerus positioned in maximum tolerated external rotation with the arm along the body and the second with the humerus in internal rotation with the palm placed flat on the table). Two independent readers blinded to the arthroscopic results evaluated the CTA images for labral tears, glenoid bone loss/fractures, and cartilage loss. For each CTA acquisition, sensitivity and specificity for detection of the aforementioned pathology were assessed. Inter-reader agreement was quantified by weighted ĸ statistics. RESULTS:Sensitivity and specificity for detecting anteroinferior or posterior labral tears was highest with neutral rotation (sensitivity 91-100%, specificity 61-100%). For glenoid fracture, sensitivity (67%) was highest with external rotation and specificity (100%) was highest with internal rotation. For cartilage loss, sensitivity (64%) and specificity (89%) was highest with external rotation and neutral rotation, respectively. Neutral rotation showed high sensitivity and specificity for glenoid fractures and cartilage loss. Inter-reader agreement ranged from fair to very good. CONCLUSIONS:Neutral glenohumeral position in shoulder CT arthrography was adequately sensitive and specific for the detection of intra-articular pathology, avoiding the use of more than one acquisition.