BACKGROUNDː Distant metastases (DM) occur rarely and are associated with poor outcomes in patients with differentiated thyroid carcinoma (DTC). The aim of this study was to explore potential risk factors of DM in DTC and build a decision-tree model. METHODSː The medical records of 15,591 patients who were diagnosed with DTC after initial surgery in three medical centers (2000 to 2018) were reviewed and 37 patients (test group) and 14 patients (validation group) with DM and detailed clinicopathologic characteristics were identified. Patients with no evidence of disease (NED) postoperatively were randomly sampled to create a control group in a 4:1 ratio. RESULTSː Multiple factors, including median age, extrathyroidal extension (ETE), AJCC stage, position, histological type, and diameter differed significantly between the DM and NED groups (P˂0.001) in univariate and multivariate analysis. AJCC stage and diameter of the primary tumor made the greatest contributions to prognosis according to decision-tree analysis and a random forest algorithm. The predictive model constructed from these data achieved 100% accuracy of classification. External validation confirmed that this model has 100% accuracy of classification. In addition, histology and ETE were found to be independent predictors of survival in patients with metachronous metastases. CONCLUSIONSː This study optimized the weight of risk factors, including AJCC stage and diameter of primary tumor, in predicting DM in patients with DTC. Our predictive model provides a strong tool for prediction that may potentially affect clinical decision-making.