The evaluation of recurrent shoulder symptoms in patients who have had rotator cuff repair is a diagnostic challenge. Pain and limitation of motion may be caused by a recurrence of the rotator cuff tear. Arthrography is not considered to be helpful in postoperative cases, since it is false-positive in most of them. Pre-operative ultrasonography of the shoulder is regarded as a highly accurate diagnostic tool for rotator cuff tears. The diagnostic criteria used are: the continuity of the rotator cuff, its contour, its thickness and its echogenicity. Our study was aimed at determining which ultrasonographic criteria were significant for a retear. In addition, the "normal" postoperative sonographic appearance of the rotator cuff was established. Of 133 patients with a full thickness rotator cuff tear, 110 were evaluated 4-48 months (mean in 21 months) after surgery. A total of 85 cuffs were reconstructed, in 46 cases by direct suture, in 33 cases by suture to the major tubercle and in 6 cases by tendon transfer (Cofield technique). In 25 cases cuffs the could not be reconstructed. Subacromial decompression was performed routinely. The patients were evaluated clinically by range of movement, force and isometric and impingement tests. The subjective outcome was assessed by the algo-functional index of Patte. Ultrasonography was performed using a 7.5-MHz linear scanner. Each sonographic criterion was referred to the clinical and subjective findings. Ultrasonographic evaluation of the rotator cuffs that could not be reconstructed revealed non-visualization of the tendons. In 13 of the 85 patients in whom reconstruction of the cuff was possible a normal sonographic pattern was seen.(ABSTRACT TRUNCATED AT 250 WORDS)