PURPOSE. We performed an anatomical study to clarify humeral insertions of coracohumeral ligament (CHL) and superior glenohumeral ligament (SGHL) and their relationship with subscapularis tendon. The purpose of our study was to explain the « Comma Sign » observed in retracted subscapularis tears treated by arthroscopy. MATERIAL AND METHODS. 20 fresh cadaveric shoulders were dissected by wide delto-pectoral approach. After removal the deltoid and posterior rotator cuff, we removed humeral head on anatomical neck. So we obtained an articular view comparable to arthroscopical posterior portal view. We looked for a structure inserted on subscapularis tendon behind SGHL. By intra-articular view we removed SGHL and CHL from the medial edge of the bicipital groove, then subscapularis tendon from lesser tuberosity. We splitted the rotators interval above the superior edge of subscapularis tendon and observed the connections between subscapularis tendon, CHL and SGHL. RESULTS. 6 shoulders had massive cuff tears and were excluded. No ligamentous structure was visible between rotators interval and subscapularis tendon by simple intra and extra-articular examination. After removal of LGHS humeral insertion, no structure showed vertical attach on tendon yet. But after removal of subscapularis tendon from lesser tuberosity and medial traction we saw constantly a fibers bundle directly inserted onto supero-lateral edge of subscapularis tendon. DISCUSSION/CONCLUSION. Most authors agree about existence of CHL and SGHL and their bone insertions, whereas relations between themselves and subscapularis tendon aren't so well defined. We constantly found an effective link between subscapularis tendon and a fibers bundle mainly coming from LCH. It layed into supero-lateral edge of subscapularis tendon and could be seen only by medial traction of it. This ligamentous structure yields the « Comma Sign » in subscapularis tendon tears. This study confirms our clinical datas