Objective:To explore the risk factors of blood glucose fluctuation and complications in patients with diabetes.Meth-ods:23 8 patients with type 2 diabetes were retrospectively selected for standardized diagnosis,treatment and follow-up.According to whether the diabetic patients had blood glucose fluctuation and diabetic complications,they were divided into blood glucose fluctuation group(n=54),non-blood glucose fluctuation group(n=184),and diabetic complication group(n=35),non-diabetic complication group(n=203).After 12-month follow-up,the risk factors of blood glucose fluctuation and complications in patients with diabetes were identified by univariate analysis and multivariate regression analysis.And establish a risk prediction model for the occurrence of blood glucose fluctuation and diabetes in patients with diabetes.Results:Univariate and multivariate regression analysis showed that male,smoking,course of disease,number of treatment lines,and higher levels of blood triglycerides(TG)were independent risk factors for causing blood glucose fluctuations in patients(P＜0.05).Age,male,smoking,course of disease,number of treatment lines,glycosylat-ed hemoglobin,higher TG and lower level of glomerular filtration rate(eGFR)were independent risk factors for diabetic complications(P＜0.05).Based on the risk factors,the risk prediction model of blood glucose fluctuation and diabetic complications in patients with diabetes were established.Through internal verification,the area under the curve calculated by ROC curve was 0.954 and 0.934,which proved that the prediction efficiency of the model was better.Conclusion:Male,smoking history,course of disease,number of treatment lines and TG level are independent risk factors for blood glucose fluctuation in patients.Age,male,smoking,course of disease,number of treatment lines,glycosylated hemoglobin,higher TG and lower eGFR levels are independent risk factors for diabetic complications.Based on the risk factors,the risk prediction model of blood glucose fluctuation and complications in patients with diabetes can be estab-lished,and the prediction efficiency of the model is good.