Objective To validate and compare the risk prediction models of gestational diabetes mellitus （GDM） in a prospective cohort study, and to provide suitable prediction tools for screening high-risk pregnant women with GDM.Methods A total of 788 pregnant women with GDM were selected into the research cohort.Eight models were used to predict the risk of GDM in the cohort, and the actual occurrence of GDM was recorded.The predictive performance of the model was evaluated by the area under the receiver operating characteristic curve （AUC）, accuracy and other indicators, and the clinical applicability of the model was evaluated by the included predictors.Results The incidence of GDM in the validation dataset was 10.2%.The AUC values of the eight models ranged from 0.54 to 0.67, and the accuracy ranged from 0.10 to 0.90.The negative predictive value was 0.10-0.89, and the positive predictive value was 0.10-0.96. Among them, Li Jinjin model was the best model with an AUC of 0.66 （95%CI 0.60, 0.72） and an accuracy of 0.80 （95%CI 0.77, 0.83）.The number of factors included in the model was appropriate and the model was easy to obtain, which had good clinical applicability.Conclusion For Chinese pregnant women, the model based on Chinese data performed well, while the model based on foreign data performed poorly.Compared with the other seven models, Li Jinjin model had better predictive performance and clinical applicability in the Chinese pregnant women cohort.