Objective: To establish a Parotid Imaging Reporting and Data System (PI-RADS) for CT diagnosis of the parotid gland neoplasms and to investigate the clinical applicable value and feasibility of PI-RADS. Methods: Patients who had been diagnosed with primary parotid gland neoplasms and had received surgical treatments in Peking University School and Hospital of Stomatology during the period of January 2013 to December 2016 were included in this study. The diagnoses were confirmed by the postoperative pathological examinations in all the patients. The CT imaging data of all patients were retrospectively reviewed and analyzed by two readers in consensus. Imaging characteristics related to the parotid neoplasms were extracted and quantified. Based on comprehensive analysis of the imaging characteristics, the probabilities of the benign and malignant neoplasms were evaluated and classified into six grades, PI-RADS 1-6 (PI-RADS 1: normal parotid gland; PI-RADS 2: confidently benign lesions; PI-RADS 3: probably benign lesions without confirmed evidence of malignancy; PI-RADS 4: suspected malignancy without sufficient evidence of malignancy; PI-RADS 5: confidently malignant lesions; PI-RADS 6: lesions with confirmed pathological evidence of malignancy). Results: A total of 897 patients with 1 003 parotid lesions were included. The lesions included 905 benign and 98 malignant lesions. The proportions of the malignancies in PI-RADS 2, PI-RADS 3, PI-RADS 4 and PI-RADS 5 according to the two readers in consensus were 0.4%, 5.7%, 35.5% and 96.7% respectively. The overall Cohen’s Kappa test showed medium consistency between the two independent researchers (κ=0.614, P<0.001, 95%CI: 0.569-0.695). Pearson Chi-square test showed that the proportions of malignancies increased with the diagnostic PI-RADS grades (Cochran-Armitage trend test, Z=-15.579, P<0.001). The results of Pearson Chi-square tests showed significant differences between the grades [PI-RADS 2 and 3 (χ ²=12.048, P=0.001); PI-RADS 3 and 4 (χ ²=75.231, P<0.001); PI-RADS 4 and 5 (χ ²=32.266, P<0.001)]. Conclusion: PI-RADS can be used to evaluate the risk of malignancy and will be helpful to improve the imaging diagnosis and clinical treatment of paro-tid gland neoplasms.