Objective To evaluate the correlation of the positron emission tomography/magnetic resonance imaging (PET/MR) parameters with the pathological differentiation of head and neck squamous cell carcinoma(HNSCC) and the diagnostic efficiencies of PET/MR parameters. Methods Patients with clinical suspicion of HNSCC were included and underwent PET/MR scan. HNSCC was pathologically confirmed in all these patients. The PET/MR examination included PET and MR sequences of diffusion-weighted imaging (DWI) and T2-and T1-weighted imaging. The multiple parameters of PET/MR included the mean values of apparent diffusion coefficient(ADC) and the maximum and mean values of standardized uptake value (SUV and SUV) were measured and estimated. The correlations of all the parameters and distribution between the different tumor differentiation groups were analyzed. Logistic regression was utilized to build the model as the PET/MR combined parameter for predicting the differentiation by multiple parameters of PET/MR. The receiver operating characteristic curve was calculated for each parameter and the combination. Results Totally 23 patients were included in this study:9 patients (9 males and 0 female) had well-differentiated tumor,with an average age of (61.0±6.8)years;14 cases had moderately-differentiated (n=10) or poorly-differentiated tumors (n=4),with an average age of (62.0±9.1) years. All the patients were males. There was statistical correlation between SUV and SUV (P<0.001);however,ADC showed no statistical correlation with SUV and with SUV (P=0.42,P=0.13). ADC and SUV showed significant difference between well-differentiated group and moderately-poorly-differentiated group (P=0.005,P=0.007). Compared with the individual parameters,the combination of PET/MR parameters with SUV and ADC had higher efficacy in predicting tumor differentiation,with an area under curve of 0.84. Conclusions The distributions of ADC,SUV and SUV differ among HNSCC with different pathological differentiation. Compared with the individual parameters,the combination of the PET/MR parameters has higher efficiency in predicting tumor differentiation.