In a retrospective study of seventy-four shoulders in seventy-three patients treated at the Geisinger Medical Center between January 1951 and March 1972, the diagnosis of tear of the rotator cuff was confirmed by surgical repair in sixty-five shoulders and by arthrography in nine. According to a described rating system, the results in the sixty-five surgically treated shoulders at the end of convalescence (from 2.8 to 21.6 months after surgery) were excellent or good in 69 per cent and fair or poor in 31 per cent, while in the forty-six shoulders reevaluated from eighteen months to twenty years after surgery the results were excellent or good in 74 per cent and fair or poor in 26 per cent. Sex, age, occupation, and a traumatic onset had little or no influence on the result. The results were better if preoperative roentgenograms were normal or if the tear was partial; while results tended to be worse if there were calcific deposits in the cuff, if the acromion was not reattached after acromionectomy, and if the repair necessitated suturing the proximal cuff into a trough cut in the humeral head. The result at six to nine months was not likely to change subsequently.