Abstract BACKGROUND Perfusion MRI (pMRI) is widely used for differentiation between progressive disease and pseudoprogression in patients with treated glioma. Although pMRI has long been available, its clinical relevance and utilization are not well known. We investigate the role of pMRI in clinical management decisions and its association with overall survival (OS) in patients with glioma under surveillance after focal radiotherapy. MATERIALS AND METHODS We performed a retrospective multicenter cohort study of patients treated for glioma between 2008 and 2021. At the moment of first substantial abnormalities on follow-up MRI, both with or without pMRI, we collected the management decisions, information about the (p)MRI and clinical parameters from the patient records. Causal inference methods were used to design a model predicting the probability of a management change with or without pMRI at first scan after radiotherapy (baseline), during chemotherapy (on-treatment) and after treatment. A multivariable Cox-proportional hazards model was used for OS analysis (adjusted for age, clinical performance, surgery type, histopathological diagnosis). Subgroup analyses were performed for patients with glioblastoma only. RESULTS 976 patients (mean age 57.2 ±12.4 years; 649 men) were included in 7 neuro-oncological referral centers (4 academic), 548 with and 428 without pMRI. Eighty percent of pMRI scans were of sufficient quality. Eighty percent of patients had glioblastoma and in this group, during the “on-treatment” phase, there was a 16% lower probability of change in clinical management in patients with pMRI, which was more pronounced (37%) in academic hospitals. OS was longer in patients with versus without pMRI with a hazard ratio of 0.77 (95%CI: 0.67,0.89, p<0.001). These findings were similar when only considering patients with glioblastoma. CONCLUSION pMRI was associated with fewer management changes while patients were on treatment, and with longer OS. This latter result must be interpreted with caution due to possible confounding.: