This review is one in a series of Cochrane Reviews of interventions for shoulder disorders. To synthesise the available evidence regarding the benefits and harms of rotator cuff repair with or without subacromial decompression in the treatment of rotator cuff tears of the shoulder. We searched the CENTRAL, MEDLINE, Embase, Clinicaltrials.gov and WHO ICRTP registry unrestricted by date or language until 8 January 2019. Randomised controlled trials (RCTs) including adults with full-thickness rotator cuff tears and assessing the effect of rotator cuff repair compared to placebo, no treatment, or any other treatment were included. As there were no trials comparing surgery with placebo, the primary comparison was rotator cuff repair with or without subacromial decompression versus non-operative treatment (exercises with or without glucocorticoid injection). Other comparisons were rotator cuff repair and acromioplasty versus rotator cuff repair alone, and rotator cuff repair and subacromial decompression versus subacromial decompression alone. Major outcomes were mean pain, shoulder function, quality of life, participant-rated global assessment of treatment success, adverse events and serious adverse events. The primary endpoint for this review was one year. We used standard methodologic procedures expected by Cochrane. We included nine trials with 1007 participants. Three trials compared rotator cuff repair with subacromial decompression followed by exercises with exercise alone. These trials included 339 participants with full-thickness rotator cuff tears diagnosed with magnetic resonance imaging (MRI) or ultrasound examination. One of the three trials also provided up to three glucocorticoid injections in the exercise group. All surgery groups received tendon repair with subacromial decompression and the postoperative exercises were similar to the exercises provided for the non-operative groups. Five trials (526 participants) compared repair with acromioplasty versus repair alone; and one trial (142 participants) compared repair with subacromial decompression versus subacromial decompression alone. The mean age of trial participants ranged between 56 and 68 years, and females comprised 29% to 56% of the participants. Symptom duration varied from a mean of 10 months up to 28 months. Two trials excluded tears with traumatic onset of symptoms. One trial defined a minimum duration of symptoms of six months and required a trial of conservative therapy before inclusion. The trials included mainly repairable full-thickness supraspinatus tears, six trials specifically excluded tears involving the subscapularis tendon. All trials were at risk of bias for several criteria, most notably due to lack of participant and personnel blinding, but also for other reasons such as unclearly reported methods of random sequence generation or allocation concealment (six trials), incomplete outcome data (three trials), selective reporting (six trials), and other biases (six trials). Our main comparison was subacromial decompression versus non-operative treatment and results are reported for the 12 month follow up. At one year, moderate-certainty evidence (downgraded for bias) from 3 trials with 258 participants indicates that surgery probably provides little or no improvement in pain; mean pain (range 0 to 10, higher scores indicate more pain) was 1.6 points with non-operative treatment and 0.87 points better (0.43 better to 1.30 better) with surgery.. Mean function (zero to 100, higher score indicating better outcome) was 72 points with non-operative treatment and 6 points better (2.43 better to 9.54 better) with surgery (3 trials; 269 participants), low-certainty evidence (downgraded for bias and imprecision). Participant-rated global success rate was 873/1000 after non-operative treatment and 943/1000 after surgery corresponding to (risk ratio (RR) 1.08, 95% confidence interval (CI) 0.96 to 1.22; low-certainty evidence (downgraded for bias and imprecision). Health-related quality of life was 57.5 points (SF-36 mental component score, 0 to 100, higher score indicating better quality of life) with non-operative treatment and 1.3 points worse (4.5 worse to 1.9 better) with surgery (1 trial; 103 participants), low-certainty evidence (downgraded for bias and imprecision). We were unable to estimate the risk of adverse events and serious adverse events as only one event was reported across the trials (very low-certainty evidence; downgraded once due to bias and twice due to very serious imprecision). At the moment, we are uncertain whether rotator cuff repair surgery provides clinically meaningful benefits to people with symptomatic tears; it may provide little or no clinically important benefits with respect to pain, function, overall quality of life or participant-rated global assessment of treatment success when compared with non-operative treatment. Surgery may not improve shoulder pain or function compared with exercises, with or without glucocorticoid injections. The trials included have methodology concerns and none included a placebo control. They included participants with mostly small degenerative tears involving the supraspinatus tendon and the conclusions of this review may not be applicable to traumatic tears, large tears involving the subscapularis tendon or young people. Furthermore, the trials did not assess if surgery could prevent arthritic changes in long-term follow-up. Further well-designed trials in this area that include a placebo-surgery control group and long follow-up are needed to further increase certainty about the effects of surgery for rotator cuff tears.