The preoperative diagnosis of thyroid nodules, which is critical for guiding therapeutic strategies, remains unsatisfactory. In this work , we investigate the use of intravoxel incoherent motion (IVIM) diffusion weighted imaging (DWI) in the differentiation between benign and malignant thyroid lesions.45 patients with 56 confirmed thyroid nodules underwent preoperative routine neck MRI and IVIM DWI. The histopathologic diagnosis was confirmed by surgery, benign (n = 40) and malignant (n = 16) nodules,postoperative nodules were performed HE staining. The DWI data was postprocessed by IVIM model for quantitation of apparent diffusion coefficient (ADC), perfusion fraction f, diffusivity D and pseudo diffusivity D*. Independent samples t test was conducted between benign nodules group and malignant nodules group on the obtained IVIM parameters. Receiver operating characteristic (ROC) analyses were conducted to determine the optimal thresholds as well as the sensitivity and specificity for differentiating.MRI scans were performed on a 3.0T MRI system (GE Discovery MR 750) with a head and neck joint coil. Conventional MRI sequences including axial T1WI (TR/TE= 488/10 ms),T2WI (TR/TE= 2400/73 ms) with 256×256 matrix size, 24 cm FOV, 1 number of excitation, 4 mm slice thickness and 1mm slice gap. Ten b values (0,30,50,80,100,200,400,600,800,1000s/mm²) were applied with a single shot diffusion weighted SE EPI sequence using parallel imaging with following parameters: FOV = 24cm ,slice thickness = 4 mm ,slice gap = 1 mm , TR/TE= 2000/55 ms,matrix= 96×96. To obtain IVIM images with sufficient SNR,NEXs of 2,4,4,2,2,2,4,6,6 and 8 corresponding to each b value were used. The total scanning time for IVIM DWI was about 7min 56s. IVIM parameters were calculated in certain ROIs of both thyroid benign and malignant nodules. The values of the IVIM parameters for cases of both thyroid benign nodules and malignant tumors are shown in Table 1. Significant intergroup difference was observed in ADC, D, D *, and f (P<0.001 ). Malignant tumors featured significantly lower ADC, D and D* values and a higher the f value than that of benign nodules. The box plots of IVIM parameters are shown in Fig. 1 of thyroid nodules clearly demonstrate these patterns. Fig. 2 shows the ROC curves of the performances of the four IVIM parameters,the ADC, D, and D * could distinguish the benign from malignant thyroid nodules, and parameter f differentiate the malignant tumors from benign nodules . The values of AUC for parameter ADC, D, and D * were 0.784, 0.795, 0.850,separately, of which the area under the curve (AUC) of f value was the maximum for identifying the malignant from benign nodules, which was 0.841, the diagnostic threshold of f value was 32.04%, and the sensitivity, specificity and accuracy was 93.8%, 70.0% and 76.7%, respectively. Fig.3 shows results of typical pair of benign nodules (a-g) and malignant ones (i-o). HE staining showed nodular goiter with varying sizes filled with gelatinous follicular structure (h), and nuclei arranged closely in papillary thyroid carcinoma (p).IVIM, first described by Le Bihan et al. ^[1] ,is an interesting imaging technique for the separately estimating microscopic perfusion and diffusivity using multiple b value DWI. In the present work ,it is suggested that several parameters of IVIM modelling, including ADC, D, D*, and f,could potentially serve as noninvasive predictors for the preoperative differentiating of thyroid nodules, and f value performed best in identifying the malignant from benign nodules among these parameters.