Abstract: The qualitative identification and quantitative prediction model of buffalo milk adulterated with water and urea were established by seven main quality indicators of the buffalo milk combined with chemo metrics methods. The cluster analysis was well able to distinguish pure and adulterated buffalo milk samples. Canonical discriminant function coefficients could intuitively distinguish mixed samples adulterated with different contents of adulteration,and coefficients of Fisher's linear discriminant functions worked well for qualitative identification of unknown buffalo milk adulteration. Multiple linear step regression method built six quantitative models,which the correlation coefficients RPof real value and predicted value were 0.9956,0.9934,0.9937,0.9938,0.9986,0.9991,respectively,and could quantitative identification of authentic buffalo milk.