Objective: To discover and prevent diabetes early in high-risk group. Methods:Chose 323 records of type 2 diabetes or IFG,IGT patients with first course which come from a large hospital in Lanzhou between 2009 January to 2014 March; Use Apriori algorithm and the SPSS Clementine 12 software design;Set up prediction model of type 2 diabetes. Results:A total of 10 strong association rules are formed. The strong association rules contain some associations between factors and type 2 diabetes. Conclusion: Through the association rules, the initial judgment on the high-risk group of diabetes is simple and accurate.