Osteoarthritis (OA) is a leading cause of functional impairment and pain. Proprioceptive defects may be associated with the onset and progression of OA of the knee. The purpose of this study was to determine the effectiveness of proprioceptive exercises for knee OA using meta-analysis. A systematic review was conducted on 12th December 2011 using published (Cochrane Library, MEDLINE, EMBASE, CINAHL, AMED, PubMed, PEDro) and unpublished/trial registry (OpenGrey, the WHO International Clinical Trials Registry Platform, Current Controlled Trials and the UK National Research Register Archive) databases. Studies were included if they were full publications of randomized or non-randomised controlled trials (RCT) comparing a proprioceptive exercise regime, against a non-proprioceptive exercise programme or non-treatment control for adults with knee OA. Methodological appraisal was performed using the PEDro checklist. Seven RCTs including 560 participants (203 males and 357 females) with a mean age of 63 years were eligible. The methodological quality of the evidence base was moderate. Compared to a non-treatment control, proprioceptive exercises significantly improved functional outcomes in people with knee OA during the first 8 weeks following commencement of their exercises (p < 0.02). When compared against a general non-proprioceptive exercise programme, proprioceptive exercises demonstrated similar outcomes, only providing superior results with respect to joint position sense-related measurements such as timed walk over uneven ground (p = 0.03) and joint position angulation error (p < 0.01). Proprioceptive exercises are efficacious in the treatment of knee OA. There is some evidence to indicate the effectiveness of proprioceptive exercises compared to general strengthening exercises in functional outcomes.