Abstract: Qualitative identification models of beef and mutton adulterated with other animals meat were established by near infrared spectroscopy( NIR) combined with principal component analysis( PCA) and discriminant analysis.The performance of the models was evaluated according to identification accuracy. Quantitative detection models of adulterated content were established by NIR combined with PCA and partial least squares( PLS). Predictive ability of the models was verified by prediction mean square error( RMSEP) and correlation coefficient( r) between the predicted values and the measured values.As a result,the accuracy of the training set and prediction set were97.86% and 91.23% respectively identified by the model of beef adulterated with pork.The accuracy of the training set and prediction set were 98.28% and 92.98% respectively identified by the model of mutton adulterated with pork.The accuracy of the training set and prediction set were 99.59% and 93.97% respectively identified by the model of mutton adulterated with duck meat.The accuracy of the training set and prediction set were 97.57% and90.76% respectively identified by the model of mutton adulteration. Interaction validation of mean square error( RMSECV) of training set samples and RMSEP of prediction set samples of beef adulterated quantitative model were 3.87% and 4.13%,and r were 0.9505 and 0.9134 respectively.RMSECV of training set samples and RMSEP of prediction set samples of mutton adulterated quantitative model were 4.48% and 4.86%,and r were 0.9306 and0.9082 respectively. The results showed that near infrared technology in combination with certain chemometrics methods can identify adulteration of animal meat from different sources,and can detect adulteration quantity.