Knee arthroscopy has historically been a common treatment for knee osteoarthritis. A Cochrane review of the literature up to 2006 has resulted in guidance that arthroscopy is not effective in knee osteoarthritis. It cited that deficiencies in the evidence base prevented widespread acceptance of the recommendations. The aim of this review is to update the evidence base for the efficacy of arthroscopy in knee osteoarthritis. The authors searched CINHAL, EMBASE, MEDLINE, and CENTRAL for randomised controlled trials that compared arthroscopic surgery in knee osteoarthritis with a control group (e.g. lavage, best medical care). The primary outcome measure was patient reported functional outcome. The study methodology was registered on Prospero, a systematic review register: Registration number CRD42012002891. Five randomised controlled trials included 516 patients, almost double the 271 episodes contained in previous reviews. Two high quality studies, according to the Jadad classification, published since the Cochrane review, addressed many of the methodological flaws criticised in previous reviews. However, certain subgroup analyses (e.g. patients with meniscal tears and mechanical symptoms) are still underpowered.