Unlabelled: Shoulder stiffness is a problem which covers many different conditions. In fact there is still a semantic and pathogenetic confusion. The words: capsulite retractile, frozen shoulder, adhesive capsulitis, stiff shoulder contracture have been successively used and this ambiguity renders the literature difficult to interpret. Moreover the cause of the stiffness which depends on the aetiology, is not always clearly known: capsular contraction, capsular adhesion, capsular scarring following trauma or surgery, extra capsular phenomenons in the subacromial bursa, muscles or tendons., Materials and Methods: 26 shoulders (25 patients) were reviewed with a follow up of 21 months using the Constant's scoring system. Patients had an average duration of symptoms for 13 months (1 to 27). Pre op passive motion was: abduction: 74 degrees, external rotation: 6 degrees, forward flexion: 84 degrees. The average motion core was 12.9/40. We distinguished three groups: primary frozen shoulder (13 cases) ; bipolar stiffness (3 cases) due to rotator cuff disease ; acquired "surgical" stiffness, (10 cases). The capsular release was performed, at the anterior rim of the glenoid fossa, purely anterior or anterior and inferior, followed by gentle manipulation. If external rotation was not improved the coraco-humeral ligament was detached from its coracoid attachment. Additional procedures were performed:acromioplasty (5 cases), bursectomy (3 cases), SLAP lesion debridement (1 case). Only 2 out 13 primary shoulders required an additional procedure., Results: 1-There were no intra-operative complications (vascular or neural). 2-Range of Motion: the average gain under anesthesia was: abduction: 72 degrees, external rotation: 34 degrees, forward flexion: 86 degrees. Final result was obtained with a mean duration of seven months. There was no difference according to the aetiology. Gain was more important in the primary group (9.69 to 34.9 vs 15.8 to 30.6). 3-Subjective results were better in the primary group. 4-Objective results demonstrated an absolute Constant's score of 70.3, that is to say 83.4 per cent of the contralateral supposed healthy shoulder. There were 3 excellent, 5 very good, 7 good, but 11 fair or poor results. The relative Constant's score was 91 per cent in the primary group and only 76 per cent in the acquired group. The difference was due to the pain and strength which were greatly improved in the primary group., Discussion: Arthroscopic release of shoulder contracture is feasible, safe and effective. For primary frozen shoulder, there is usually spontaneous recovery. Indications for surgery are very few. There is no evidence that arthroscopic release shortens spontaneous evolution. Therefore, we propose it in very selected cases of dramatically limited motion. One year of evolution is an acceptable time. For bipolar stiffnesses, arthroscopy allows one to recognize the exact cause of the stiffness and to treat it, especially the subacromial pathology. In this occurrence, buroscopy must be performed and cuff pathology treated. For acquired surgical stiffnesses, gain of motion is significant. Subjective and objective results are less satisfactory than those of primary frozen shoulder, due to persistance of pain and lack of strength. The alternative is open release, but arthroscopic release has less morbidity. It can be proposed early as soon as capsular tissue has healed (for instance 6 months).