A clinical study of 200 unselected elderly subjects with an average age of 78 years living in retirement homes was performed to determine the prevalence of clinically manifest rotator cuff tears and the relevance of the tear for daily life. A comprehensive questionnaire was performed and the examination included the UCLA- and Constant-shoulder assessment scales and a multitude of special tests. The criteria for clinical diagnosis of a rotator cuff tear were a positive Jobe-test, diminished external rotation, a positive zero-degree-abduction test, the drop arm sign, pseudoparalysis and atrophy. By these criteria we found 28 (7%) rotator cuff tears in 25 people. There were 15 women and 10 men and the mean age of symptom onset was 55 years. Eleven craftsmen, 7 office workers and 7 housewives met our criteria for rotator cuff tears. Pain was present in 22 shoulders. Moderate weakness was found in 2 people and severe weakness in 22. In 20 people the range of motion was severely reduced and in 8 moderately reduced. Activities of daily living were impaired in all affected subjects. Eight people reported onset of symptoms after trauma. Patients with clinically diagnosed rotator cuff tears scored an average of 13 points in the UCLA-shoulder assessment scales and 40 points in the Constant-shoulder assessment scales. In conclusion rotator cuff may be symptomatic only in some people. Probably more anatomic than symptomatic rotator cuff tears exist, and it can be supposed that physical activities may be one possible reason for clinically relevant tears.