Purpose: Abnormal scapular motion, dyskinesis, increases the risk of shoulder injury in athletes. Previous research identified alterations in 3D scapular kinematics in overhead athletes with dyskinesis during concentric elevation. However, clinically dyskinesis is observed during the eccentric phase of elevation. The discrepancy between clinical observations and kinematic research may be related to the inconsistency, or variability, in scapular motion during eccentric elevation. The purpose of this study is to compare the variability of scapular movement between phases of elevation and groups, overhead athletes with dyskinesis (DYSK) and without dyskinesis (w/o DYSK). Subjects: 12 overhead athletes participated (N=7 w/o DYSK; N=5 DYSK, matched by age, arm dominance, and gender) Methods: Scapular motion was evaluated with the scapular dyskinesis test. Three-dimensional shoulder kinematics were collected while subjects performed 10 repetitions of weighted bilateral flexion. Mean scapular kinematic data for upward/downward rotation (UR/DR), internal/external rotation (IR/ER), and anterior/posterior (AP) tilt were computed. Scapulohumeral joint coupling variability was calculated with a vector coding technique that quantifies the variability in vector angles. To determine differences in scapular kinematics and joint coupling variability, a curve analysis was used. Results: There were significant differences in scapular AP tilt kinematics and variability between groups and phases. Greater variability in AP tilt was found during the eccentric phase in DYSK. Conclusions: Increased variability suggests alterations in scapular kinematics exist during the eccentric phase of elevation. Clinical Relevance: This study suggests clinicians should examine scapulohumeral rhythm with a focus on variable movement during the eccentric phase.