Rotator cuff tears (RCT) are frequent and increase with age. Why do only a relatively small percentage cause a permanent severe handicap justifying surgery? Is there a relationship between size and site of the tear and the clinical symptoms? How do they influence the postoperative result? These questions were answered in a prospective study of 76 operated patients. We adopted D. Patte's classification of the RCT into four groups according to the site and size of the lesion. Our figures were compared with those obtained by Patte in an analogous study of 256 cases. Of the tears in our series, 56.5% belonged to groups I and II, which means that only the supraspinatous and occasionally also the subscapular muscles were involved. In 100% of the cases in group I we found only pain, whereas a pure symptomatology with pain alone was presented in only 51% of group II, in 48% of group III, and in 0% of group IV. Groups III (35.5%) and IV (8%) represent more extensive tears, involving not only the anterior part of the rotator cuff, e.g., the supraspinatous tendon, but also to a more or less severe extent, the infraspinatous and sometimes even the teres minor. In group IV the extensive tear is combined with osteoarthritic changes. In both groups a mixed symptomatology (pain and pseudoparalysis) is the rule. The result of operative treatment is satisfactory in all groups as far as pain relief is concerned. The improvement of active ROM is not as evident and seems, as expected, to be related to the severity of the lesion.(ABSTRACT TRUNCATED AT 250 WORDS)