Experience with rotator cuff sonography in 106 patients was analyzed to identify the causes of scan misinterpretation. Possible interpretive errors may be classified into four categories: those from failure to recognize normal anatomy, those caused by technical limitations of the study. Errors in recognition of normal anatomy are easily overcome by experience and comparison to the normal, contralateral rotator cuff. Errors resulting from soft-tissue abnormalities were seen in two patients with calcific tendinitis simulating rotator cuff tears. Problems in interpretation resulting from fractures in two patients and inferior glenohumeral subluxation in two other patients could have been avoided by review of the plain films before performing sonography. The major technical limitation of the study arises from the inability to image the rotator cuff beneath the acromion. Fortunately, rotator cuff tears are rarely isolated in this location, and passive maneuvers often allow otherwise hidden parts of the cuff to be imaged.