Abstract: The calibration analysis of moisture,caffeine and tea-polyphenols were studied by combined near infrared spectroscopy and partial least squares (PLS),based on in random selection of 42 instant teas. The evaluation index of prediction model included correlation coefficient of calibration (R_c),the root mean square error of prediction (RMSEC),cross-validation correlation coefficient of calibration (R_cv),and the root mean square error of cross-validation error (RMSECV). The results showed that the ideal calibration model of Rc,RMSEC,R_cv,and RMSECV for moisture were 0.9266,0.6439,0.8809,0.8509.The optimal effect of calibation model for caffeine were 0.9964,0.1337,0.9543,0.4958.The ideal modeling of the R_c,RMSEC,R_cv,and RMSECV for tea-poly phenols were 0.9845,1.2097,0.9679,1.8083.The prediction correlation coefficient of moisture,caffeine and tea polyphenols was 0.9050,0.9350 and 0.9557 respectively after the validation set sample test,which has a high agreement with the actual measured value and can be used as a reference for the rapid detection of the physicochemical components of the instant Pu'er tea products.