Objective To establish a hypoglycemia risk prediction model for type 2 diabetic patients, and to verify its predictive effect. Methods The related data of 1149 type 2 diabetic patients were selected using convenience sampling, which were then randomized into a modeling group （n=766） and a validation group （n=383）.The logistic regression analysis was used to establish the prediction model, the Hosmer-Lemeshow test and the area under the ROC curve were utilized to judge the goodness of fit and predictive effect of the model. Results The occurrence rate of hypoglycemia in the modeling group and validation group was 26.76% and 23.24% respectively. The logistic regression analysis showed that, body mass index （OR=0.120）, length of hospital stay （OR=2.052）, duration of diabetes （OR=6.434）, peripheral vascular disease （OR=2.979）, glycosylated hemoglobin （OR=0.215） and triacylglycerol （OR=0.470） were the main influencing factors of hypoglycemia. As for the modeling group, the area under the ROC curve was 0.867, with the maximum value of Youden index was 0.614, the optimal critical value was 0.329, the sensitivity was 0.766, the specificity was 0.848, and H-L test P=0.071.The test results of the validation group data indicated that, the area under the ROC curve was 0.895, the sensitivity was 0.730, and specificity was 0.884. Conclusion The established prediction model has good predictive effect, which could provide reference for clinical medical staff identifying the high-risk groups of hypoglycemia early.