Seven quantitative analysis models for Dafo Longjing tea, including tea color, liquor color, aroma, taste, infused leaf, total points of five factors and total points of six factors, were built by applying near infrared spectroscopy combined with partial least squares (NIRS-PLS), in order to find a objective and scientific method for tea quality evaluation. Results showed that both the calibration samples and the prediction samples of the seven models had acquired a high fitting degree when the number of principal components was below 10, and the value of Rc, RMSEC, Rp and RMSEP were between 90.48%～98.43%, 1.14%～2.09%, 90.00%～96.65%, and 1.52%～2.84%, respectively. Among them, the total points of five factors model had the best prediction performance, and the value of Rp and RMSEP was 96.65% and 1.52%, respectively. Moreover, it was also found that the prediction precision of total points models were higher than that of single factor ones. It seems that the quality evaluation of Dafo Longjing tea could be realized by using NIRS-PLS to some extent.