Background/aim: Studies have shown that mental imagery can enhance relearning and generalisation of function after stroke. The aim of this meta-analysis was to evaluate evidence on the effects of mental imagery on motor recovery of the hemiplegic upper extremities after stroke. Methods: A comprehensive data base search of the literature up to December 2012 was performed using PubMed, EBSCO host (Academic Search Premier, CINAHL and Educational Resource Information Center), PsycINFO, Medline, and ISI Web of Knowledge (Science Citation Index and Social Sciences Citation Index). Randomised clinical trials or controlled clinical trials that included mental imagery for improving upper extremity motor function for stroke patients were located. Relevant articles were critically reviewed and methodological quality was evaluated using the PEDro Scale, and study results synthesised. Results: Five randomised clinical trials and one controlled clinical trial met the inclusion criteria. Five of the six studies yielded positive findings in favour of mental imagery. Quantitative analysis showed a significant difference in the Action Research Arm Test (overall effect: Z=6.75; P<<0.001). Conclusion: Review of the literature revealed a trend in support of the use of motor imagery for upper extremity motor rehabilitation after stroke. Mental imagery could be a viable intervention for stroke patients given its benefits of being safe, cost-effective and rendering multiple and unlimited practice opportunities. It is recommended that researchers incorporate imaging techniques into clinical studies so that the mechanism whereby mental imagery mediates motor recovery or neural adaptation for people with stroke can be better understood.