Question: In patients with small or medium-size rotator cuff tears, how do surgery and physiotherapy compare? Design: Randomized (allocation concealed), blinded (outcome assessor) controlled trial with 12-month follow-up. Setting: A hospital in Baerum, Norway. Patients: 103 patients (mean age 60 y, 71 % men) with small (<1 cm) or medium-size (1 to 3 cm) tears of the rotator cuff. Inclusion criteria were lateral shoulder pain at rest or with exercise, a painful arc of shoulder motion, positive impingement signs, and a passive range of movement of ≥140° for abduction and flexion. Exclusion criteria included age of < 18 years, tears absolutely requiring surgery, concomitant local or systemic diseases affecting shoulder function, previous tendon surgery on the relevant shoulder, and medical comorbid conditions. Follow-up data were available for all patients. Intervention: Patients were allocated to surgery (n = 52) or physiotherapy (n = 51). Surgery involved mini-open oropen tendon repair. Following diagnostic arthroscopy and a deltoid-splitting approach, an anteroinferior acromioplasty was performed. The deltoid was repaired to the acromion through drill-holes. Patients in the physiotherapy group received outpatient treatment by 1 of 4 physiotherapists. 40-minute sessions were given on the average of twice weekly for 12 weeks, and thereafter at increasing intervals during the next 6 to 12 weeks. Exercises to center the humeral head in the glenoid fossa were done to achieve glenohumeral control, and then progressed to more challenging positions. Additional exercises were given, tailored to specific demands in work, sports, and leisure activities. Patients who did not improve after 15 sessions were reassessed and offered surgery if deemed necessary. Main outcome measures: The primary outcome measure was an assessment of shoulder function with use of the Constant score (range, 0 to 100; higher scores represent better function). Secondary outcomes included the self-report section of the American Shoulder and Elbow Surgeons (ASES) score (range, 0 to 100); the Short Form (SF)-36 (0 to 100 points; higher scores indicate better health condition); Constant subscores on pain-free active shoulder abduction, pain-free active shoulder flexion, and shoulder strength; and ASES subscore on shoulder pain. Main results: Analysis was by intention to treat. 9 patients (18%) in the physiotherapy group showed no improvement after a mean of 24 sessions and underwent secondary surgical treatment. Missing nonoperative follow-up data of these patients after secondary surgical treatment were handled by a last-observation-carried-forward principle, meaning that the last scoring result before secondary surgical treatment was used for analysis. Differences between the surgery and physiotherapy groups on the Constant and ASES scores showed greater improvement from baseline in the surgery group than in the physiotherapy group (Table). Among the subscores, the surgery group patients had larger increases in shoulder abduction and decreases in pain; groups did not differ with regard to shoulder flexion or shoulder strength (Table). The groups were not different with regard to SF-36 scores. Conclusion: In patients with small or medium-size rotator cuff tears, 1-year outcomes were better after surgery than after physiotherapy.