We have compared three different methods of treating symptomatic non-traumatic tears of the supraspinatus tendon in patients above 55 years of age. A total of 180 shoulders (173 patients) with supraspinatus tendon tears were randomly allocated into one of three groups (each of 60 shoulders); physiotherapy (group 1), acromioplasty and physiotherapy (group 2) and rotator cuff repair, acromioplasty and physiotherapy (group 3). The Constant score was assessed and followed up by an independent observer pre-operatively and at three, six and twelve months after the intervention. Of these, 167 shoulders were available for assessment at one year (follow-up rate of 92.8%). There were 55 shoulders in group 1 (24 in males and 31 in females, mean age 65 years (55 to 79)), 57 in group 2 (29 male and 28 female, mean age 65 years (55 to 79)) and 55 shoulders in group 3 (26 male and 29 female, mean age 65 years (55 to 81)). There were no between-group differences in the Constant score at final follow-up: 74.1 (sd 14.2), 77.2 (sd 13.0) and 77.9 (sd 12.1) in groups 1, 2 and 3, respectively (p = 0.34). The mean change in the Constant score was 17.0, 17.5, and 19.8, respectively (p = 0.34). These results suggest that at one-year follow-up, operative treatment is no better than conservative treatment with regard to non-traumatic supraspinatus tears, and that conservative treatment should be considered as the primary method of treatment for this condition.