The objective was to describe the imaging appearances and location of delamination tears of the rotator cuff tendons on non-contrast conventional MR imaging. This study was reviewed and approved by our Institutional Review Board. The reports of 548 consecutive MR examinations of the shoulder were reviewed, looking for mention or description of delamination tears of the rotator cuff. The images of the identified cases were then reviewed by two radiologists to confirm the findings. Correlation with surgical and arthroscopic information was then performed. Delamination tears were defined as horizontal retraction of either the bursal or articular surface of the tendon, manifest as thickening of the torn retracted edge, and/or interstitial splitting of the tendon, manifest as fluid-like high signal intensity on fat-suppressed T2-weighted oblique coronal images. Fourteen cases of delamination tears were identified in 13 patients. Ten of the cases involved the supraspinatus tendon, all with articular surface involvement. Nine of these supraspinatus cases were isolated tears and one occurred as part of a full thickness tear. All 10 of these supraspinatus cases showed medial retraction of the articular surface of the tendon, with thickening of the retracted edge, and 5 of the 10 had a demonstrable horizontal cleft in the interstitium. Four cases involved the subscapularis tendon, with articular surface disruption in three and pure interstitial delamination in one. Medial subluxation of the tendon of the long head of the biceps was present in all four cases. No delamination tears occurred on the bursal surface. Only three of the 14 shoulders underwent surgical repair with one confirmation of supraspinatus delamination, one confirmation of a subscapularis tear that had become a full thickness tear 10 months after initial imaging and another interstitial subscapularis delamination that was not identified arthroscopically. Delamination tears occur most often in the supraspinatus tendon, and in our series always involved the articular surface of the tendon. Only half of these tears have a visible cleavage plane in the interstitium, but this cleavage plane is not necessary for diagnosis if the torn edge is retracted.