Although shoulder kinematics have been analyzed by various methods, dynamic shoulder motion is difficult to track. This study aimed to validate the shoulder kinematic analysis using upright four-dimensional computed tomography (4DCT) and to compare the results with optical three-dimensional motion capture. During active elevation, bilateral shoulders of 10 healthy volunteers were tracked using 4DCT and motion capture. The scapulothoracic and glenohumeral rotations and the scapulohumeral rhythm (SHR) at each position were calculated, and the differences between 4DCT and motion capture were compared. During 10−140° of humerothoracic elevation, the scapulothoracic joint showed upward rotation, internal rotation, and posterior tilting, and the glenohumeral joint showed elevation, external rotation, and anterior plane of elevation in both analyses. In scapulothoracic rotations, the mean differences between the two analyses were −2.6° in upward rotation, 13.9° in internal rotation, and 6.4° in posterior tilting, and became significant with humerothoracic elevation ≥110° in upward rotation, ≥50° in internal rotation, and ≥100° in posterior tilting. In glenohumeral rotations, the mean differences were 3.7° in elevation, 9.1° in internal rotation, and −8.8° in anterior plane of elevation, and became significant with humerothoracic elevation ≥110° in elevation, ≥90° in internal rotation, and ≥100° in anterior plane of elevation. The mean overall SHRs were 1.8 in 4DCT and 2.4 in motion capture, and the differences became significant with humerothoracic elevation ≥100°. The 4DCT analysis of in vivo shoulder kinematics using upright computed tomography scanner is feasible, but the values were different from those by skin-based analysis at the elevated arm positions.