A prospective study was done on 73 patients who underwent surgical repair of a rotator cuff tear. The patients were divided into four groups according to the active shoulder flexion and the trophism of the rotator cuff muscles: group A--range of motion > 100 degrees; group B--ROM 60-100 degrees; group C--ROM < 60 degrees; group D--ROM < 100 degrees and significant muscle hypotrophy. The rotator cuff tears were classified into four groups at surgery according to size: grade I measured < 2 cm; grade II measured 2-4 cm, grade III measured > 4 cm but was reparable; and grade IV was irreparable. The results were evaluated after an average follow-up of 2.3 years and correlated to several factors including the preoperative clinical assessment, the size of the lesion, the type of lesion (tear or detachment), and the mechanism of injury (traumatic or atraumatic). Seventy-three percent of the cases had satisfactory results. The preoperative clinical assessment and the size of the tear were the most important indicators of the final outcome. The proportion of satisfactory results underwent a progressive decline from group A (88%) to group D (14%) and from grades I and II (88% and 89%, respectively) to grade III (56%) and grade IV (none). Rotator cuff repair is almost always successful in patients with more than 60 degrees of active arm flexion and either small or medium-size tears. Less than two-thirds of the patients with major tears and less than 60 degrees of motion achieve satisfactory results. Failure is highly probable in irreparable tears and in the presence of significant hypotrophy of the rotator cuff muscles.