Objective To investigate the value of the dynamic susceptibility contrast enhanced MRI (DSC-MRI) in differential diagnosis of glioblastoma,solitary cerebral metastatic tumors and cerebral lymphoma.Methods Seventeen patients with glioblastoma,15 cases with solitary cerebral metastatic tumor and 17 cases with cerebral lymphoma were analyzed retrospectively.All patients underwent conventional MR imaging,contrast enhancement and DSC-MRI preoperatively.Pseudo color pictures of cerebral blood volume (CBV) and the time signal intensity curve were obtained from the raw data of DSC MRI.The relative CBV (rCBV)were measured from regions of enhanced solid parts of the tumors,peritumoral region and contralateral normal white matter regions respectively.The percentage of signal intensity recovery (PSR) of enhanced solid parts of the tumors were measured.ROC curve analysis was performed to determine optimum indicator in differential diagnosis of three types of tumors,and the sensitivity and specificity were calculated.Results Three types of tumors all showed enhancement of solid area with obvious peritumoral edema.Besides the no difference between glioblastoma and metastasis in rCBV of solid parts of the tumors,there were statistically significant differences in comparisons of two types of tumors (all P＜0.05).Besides the no difference between single brain metastases and lymphoma in rCBV of peritumoral regions,there were statistically significant differences in comparisons of two types of tumors (all P＜0.05).The PSR of the solid parts of the tumors had no difference between glioblastoma and single brain metastases,while there were statistically significant differences in comparisons of two types of tumors (all P＜0.05).ROC curve analysis showed sensitivity and specificity of the PSR values of solid parts of the tumors in differentiating lymphoma and non lymphoma were 100 ％and 81.3 ％.The rCBV of peritumoral regions was the optimum indicator for differentiating glioblastoma and solitary brain metastasis,the sensitivity and specificity were respectively 94.1％ and 86.7％.Conclusion The combination of rCBV and PSR can improve the efficiency for diagnosing the three types of brain tumors.