Abstract Background Arterial hyperoxia may induce vasoconstriction and reduce cardiac output, which is particularly undesirable in patients who already have compromised perfusion of vital organs. Due to the inaccessibility of vital organs in humans, vasoconstrictive effects of hyperoxia have primarily been studied in animal models. However, the results of these studies vary substantially. Here, we investigate the variation in magnitude of the hyperoxia effect among studies and explore possible sources of heterogeneity, such as vascular region and animal species. Method Pubmed and Embase were searched for eligible studies up to November 2017. In vivo and ex vivo animal studies reporting on vascular tone changes induced by local or systemic normobaric hyperoxia were included. Experiments with co-interventions (e.g. disease or endothelium removal) or studies focusing on lung, brain or fetal vasculature or the ductus arteriosus were not included. We extracted data pertaining to species, vascular region, blood vessel characteristics and method of hyperoxia induction. Overall effect sizes were estimated with a standardized mean difference (SMD) random effects model. Results We identified a total of 60 studies, which reported data on 67 in vivo and 18 ex vivo experiments. In the in vivo studies, hyperoxia caused vasoconstriction with an SMD of − 1.42 (95% CI − 1.65 to − 1.19). Ex vivo, the overall effect size was SMD − 0.56 (95% CI − 1.09 to − 0.03). Between-study heterogeneity (I 2) was high for in vivo (72%, 95% CI 62 to 85%) and ex vivo studies (86%, 95% CI 78 to 98%). In vivo, in comparison to the overall effect size, hyperoxic vasoconstriction was less pronounced in the intestines and skin (P = 0.03) but enhanced in the cremaster muscle region (P