Purpose To investigate the value of magnetic resonance imaging (MRI) features for prediction of tumor aggressiveness in papillary thyroid cancer (PTC). Method A consecutive series of 119 lesions with PTC were preoperatively evaluated by MR imaging. A multivariate analysis was performed to predict tumor aggressiveness by 18 independent variables. Results Multivariate analysis demonstrated that lesion size classification and tumor margin on delayed contrast-enhanced images were the independent predictors of high aggressive PTC. Conclusion Lesion size classification and tumor margin on delayed contrast-enhanced images can be used preoperatively to estimate the risk of high aggressive of PTC.