The contents of protein and fat in dairy products are determined quickl y by the near infrared transmittance spectroscopy (NITS).The spectroscopies are preprocessed by the t wo mixed methods:the second derivative adding S-G smoothing (SD+S-G) and the first derivative adding multi plicative scatter correction adding S-G smoothing (FD+MSC+S-G),respectively,and thenthe processed spectroscopic d ata are compressed by the wavelet with function db3and compression level 6.The quantitative analysis models of pr otein and fat in the four dairy products are established by radial basis function artificial neural network (RB F-ANN) using the compressed spectroscopy data as the input variables.The best spread value is obtained b y repeated experiment.When the spread is 135,the prediction accuracy of protein is the hig hest and the correlation coefficient and mean square error are 0.9999and 0.0301,respectively.In the same way,when the spread is 105,the prediction accuracy of fat is the highest and the correlation coefficient and mean square error are 0.9997and 0.0968,respectively.The re sults show that the model based on RBF-ANN combined with wavelet is more stable and with a higher accuracy.It could be used to test the q ualities of dairy products quickly and destructively.