Edaravone has been used in patients with acute ischemic stroke in Japan for over 10 years but does not have marketing authorization in Europe or America. Either patients in Europe and America are not receiving an effective treatment, or those in Asia are being given a treatment which is not effective. Finding out which of these is true will require further clinical trials, and a better understanding of its efficacy in animal models may help inform the design of those trials so that it might be tested under conditions where there is the greatest prospect of success. We systematically reviewed the efficacy of edaravone in animal models of focal ischemia and summarized data using weighted mean difference DerSimonian and Laird random-effects modeling. We used stratified meta-analysis and metaregression to assess the influence of study design and methodological quality. We identified 49 experiments describing outcome in 814 animals; 30 experiments (519 animals) reported functional and 35 experiments (503 animals) reported structural outcome. Edaravone improved functional and structural outcome by 30·3% (95% confidence interval 23·4–37·2%) and 25·5% (95% confidence interval, 21·1–29·9%), respectively. For functional outcome, there was an inverse relationship between study quality and effect size (P < 0·0017). Effect sizes were larger in studies where randomization or blinded assessment was not reported. There was no evidence of publication bias. Edaravone is a promising treatment for stroke. However, because of the methodological weakness in current animal studies, no sufficient preclinical evidence is available to optimize the study design of clinical trials. Higher quality animal studies are expected to inform further clinical study.