Abstract:This study was conducted to investigate the 224 samples from 4 different brands and their adulterated milk which based on the near-infrared diffuse reflection spectrum. Each sample was loaded 3 repetitions and then collected its spectrum. The principal component analysis （PCA） was applied to build the clustering discriminan model which could identificate the adulteration milk and the different brands. It turned out that the cumulative contribution rate of fitting model reached 98.98 % when the number of principal components was 3. The fitting model can correctly identify the adulterated milk and the different brands of milk ， meanwhile the method is convenient， rapid and accurate. It can provide the feasibility basis that the NIR technology is applicated in the milk quality supervision.