Bevacizumab (BVZ) is an anti-angiogenic agent for treatment of recurrent malignant glioma. In many clinical trials and practice, perfusion MRI either dynamic susceptibility-contrast (DSC)-MRI and dynamic contrast-enhanced (DCE)-MRI were used as a pharmacodynamic biomarker (i.e., to evaluate drugs’ anti-angiogenic effect on tumors) and as a predictive/prognostic biomarker [i.e., to predict progression free survival (PFS) and overall survival (OS)].1,2 We aim to systematically review the published literature and determine the value of perfusion MRI as predictive/prognostic biomarkers and pharmacodynamic biomarkers in patients with recurrent malignant glioma treated with BVZ.We identified literature that investigated perfusion MRI to predict patients’ outcome and to analyze anti-angiogenic effect of BVZ on recurrent glioma by performing a systematic search of MEDLINE and EMBASE. Study quality was assessed using REMARK guidelines. For studies reporting time-to-event data of PFS or OS, meta-analysis was performed to generate pooled hazard ratio (HR). For studies reporting pharmacodynamic change of perfusion MRI parameters, a qualitative summary was performed. To further explore study heterogeneity, subgroup analysis was performed. Of 222 articles screened, we found 13 eligible studies which had overall good quality. In our meta-analysis evaluating the predictive value of rCBV of DSC-MRI based on five studies, the pooled HR comparing PFS between responders and non-responders was 0.46 (95% CI, 0.28–0.76) (Fig. 1). In the meta-analysis for prognostic value of rCBV of DSC-MRI based on six studies, the pooled HR comparing OS was 0.47 (95% CI, 0.29–0.76) (Fig. 2). These imply that rCBV is predictive of PFS and OS after BVZ treatment. There was substantial heterogeneity in these meta-analyses, which was mainly attributed from different rCBV biomarkers. In 9 studies reporting pharmacodynamic change of perfusion MRI parameters, most perfusion MRI parameters demonstrated a consistent decrease on the follow-up MRI after BVZ treatment. The current evidence in the literature shows that the rCBV of DSC-MRI is the most widely used perfusion MRI parameter. The rCBV can be used to predict disease progression and overall survival in patients with recurrent malignant glioma treated with BVZ. Various perfusion MRI parameters from DSC-MRI and DCE-MRI could play a role as pharmacodynamic biomarkers to evaluate the drug’s anti-angiogenic effect on tumor, but standardization of the imaging acquisition and analysis techniques is necessary. Despite these unsolved issues, the current evidence favoring utilization of perfusion MRI should be considered in the clinical trials and practice for patients with recurrent malignant glioma treated with BVZ.