Objective: To investigate the diagnostic value of multimodal magnetic resonance imaging (MRI) in primary central nervous system lymphoma (PCNSL). Methods: The multimodal MRI findings of 16 patients with PCNSL confirmed by histopathology from January 2016 to December 2018 in Affiliated Hospital of North Sichuan Medical College were retrospectively analyzed. Routine MRI plain scan, enhancement, diffusion weighted imaging (DWI) and magnetic resonance spectroscopy (MRS) examination were performed in all cases. Results: All 16 patients were diagnosed with diffuse large B cell non-Hodgkin lymphoma. There were 7 cases with solitary lesions and 9 cases with multiple lesions. A total of 36 lesions were detected, with 30 above and 6 below the curtain. Sixteen lesions were located in the deep white matter near the ventricle (among which 2 lesions involved the corpus callosum), 5 lesions in the superficial cerebral hemisphere, 7 lesions in basal ganglia, 2 lesions in thalamus, 3 lesions in cerebellum, and 3 lesions were in brainstem. T₁WI showed equal or slightly low signal, and T₂WI showed equal or slightly high signal. There was different degree of edema around the lesions. 32 lesions were mild to moderate, and 4 lesions were severe. On enhancement scanning, 31 lesions showed patchy, nodular or lumpy uniform enhancement; 5 lesions showed irregular annular enhancement. "Incision" was found in 12 lesions, and "angular" was found in 7 lesions. Obvious enhancement of adjacent meninges or ependymal membranes occurred in 1 case each. DWI showed 29 high signals in the focal areas and 7 equal or slightly high signals, and all apparent dispersion coefficient (ADC) values were reduced. MRS showed that choline peak increased, and N-acetyl aspartate (NAA) peak and creatine peak decreased in all cases. highly elevated lipid peak was observed in 8 cases, inverted lactic acid peak in 3 cases, and both lipid peak and lactic acid peak were observed in 2 cases. Conclusions Multimodal MRI can reflect the characteristics of PCNSL from different aspects, which is of great value in improving the diagnostic accuracy of PCNSL.