Purpose: To demonstrate the capability of the fractional motion (FM) model for describing anomalous diffusion in cerebral gliomas and to assess the potential feasibility of FM for grading these tumors. Methods: Diffusion MRI images were acquired from brain tumor patients using a special Stejskal-Tanner diffusion sequence with variable diffusion gradient amplitudes and separation times. Patients with histopathologically confirmed gliomas, including astrocytic and oligoastrocytic tumors, were selected. The FM-related parameters, including the Noah exponent (a), the Hurst exponent (H), and the memory parameter (mu = H - 1/alpha), were calculated and compared between low-and high-grade gliomas using a two-sample t-test. The grading performance was evaluated using the receiver operating characteristic analysis. Results: Twenty-two patients were included in the present study. The calculated a, H, and m permitted the separation of tumor lesions from surrounding normal tissues in parameter maps and helped differentiate glioma grades. Moreover, a showed greater sensitivity and specificity in distinguishing low-and high-grade gliomas compared with the apparent diffusion coefficient. Conclusion: The FM model could improve the diagnostic accuracy in differentiating low-and high-grade gliomas. This improved diffusion model may facilitate future studies of neuropathological changes in clinical populations. (C) 2017 International Society for Magnetic Resonance in Medicine.