Objective: The purpose of this review was to investigate whether motor imagery training has an effect on the recovery of upper extremity function in individuals with hemiparetic stroke or not. Design: A systematic review and meta-analysis. Methods: PubMed and three other databases were searched up to December 18th, 2018 and randomized controlled trials (RCTs) evaluating motor imagery training on upper extremity function in persons with a diagnosis of hemiparetic stroke were included. This review selected the following information from each study: surname of the first author, published year, nation, population, intervention, therapeutic intensity of intervention, therapeutic comparison, outcome measures, additional therapy, summary of results, and descriptive statistics of outcome measures. Results: This review selected seventeen RCTs with 487 stroke survivors and the following intervention methods: six motor imagery training with additional therapeutic technology, two motor imagery training with additional modified constraint-induced therapy, four mirror therapy, and five motor imagery training. Ten RCTs were eligible for meta-analysis after systematic review. The motor imagery group were more effective than the control group based on the Fugl-Meyer assessment (3.43; 95% confidence interval [CI], 1.65 to 5.22; heterogeneity [$chi^2=8.03$, df=8, $I^2=0%$], test of overall effect Z=3.76; test for subgroup differences [$chi^2=2.56$, df=2, $I^2=21.8%$]) and the Action Research Arm Test (1.32; 95% CI, -8.12 to 10.76; heterogeneity [$Tau^2=70.74$, $chi^2=15.22$, df=3, $I^2=80%$], test of overall effect Z=3.76). Conclusions: The results of this review suggests that motor imagery shows positive effectiveness on improving upper extremity function in persons with hemiparetic stroke.