【Objective】We assessed the associations between protein intake and T2D in different dietary patterns. 【Methods】We conducted a population-based cross sectional study of 5 956 women and 2 298 men in middle-aged and elderly population in the community of Guangzhou city，Guangdong province. Cluster analysis was used to category subjects into mutually exclusive group by major sources of protein. Logistics regression analysis and substitution models were used to estimate T2D risks according to protein intake. 【Results】The mean age of subjects was 55.51 year. Subjects were divided into four dietary protein food patterns（coarse cereals，red meat，refined grain，and sea food）. Overall， extreme quantile of total protein intake was significantly and positively associated with T2D［odds ratio 1.58（95%CI 1.25，2.02）］. In subgroup analysis by dietary patterns，extreme quantile of total protein intake was also positively related to T2D in the“red meat”［3.00（1.36，7.01）］and“refined grain”［2.25（1.40，3.61）］dietary pattern group. However，this association was revered in the“coarse cereals”group［0.48（0.23，0.97）］；and protein intake was not related to T2D in the“seafood”group. 【Conclusions】The association between protein intake and T2D varies by dietary pattern. Dietary pattern should be considered into the recommendation of protein intake for diabetes prevention.