Cochrane Controlled Trials Register, MEDLINE, EMBASE, CINAHL and Science Citation Index databases were searched in March 2013 unrestricted by date or language.

Controlled randomized (RCT) or quasi-randomized clinical trials comparing surgery and conservative treatment of shoulder impingement were included.

The methodological quality of each included trial was assessed according to the Cochrane Collaboration's domain-based evaluation framework.

Of seven included RCTs, risk of systematic bias was considered to be low for two, high for four, and unclear for one RCT. The random-effect meta-analysis was conducted on four RCTs involving 347 subjects (173 surgically treated cases and 174 controls). There was no significant difference in changes in pain intensity between surgically and conservatively treated subjects (Hedges's g = 0.01 in favor of conservative treatment, 95% CI -0.27 to 0.30).

Based on the review of seven RCTs, the evidence on effectiveness of surgical or conservative treatment of shoulder impingement was found to be limited. There was, however, moderate evidence that surgical treatment is not more effective than active exercises on reducing pain intensity caused by shoulder impingement. Implications for Rehabilitation Based on the review of seven RCTs, the evidence on effectiveness of surgical or conservative treatment of shoulder impingement was found to be limited. There was moderate evidence that surgical treatment is not more effective than active exercises on reducing pain intensity caused by shoulder impingement. Because of surgery's higher costs and susceptibility for complications compared with costs and risks of conservative treatment, conservative treatment can be recommended as a first choice of treatment of shoulder impingement in first or second grade.