Abstract: According to electronic nose principle and aroma characteristic of Xihu-Longjing tea, the method of different tea grading discrimination based on electronic nose was investigated. The aroma of tea was prepared by mixing the tea and water in headspace vials, which could increase the respond information of electronic nose and decreasing the error of making tea. The parameters of headspace sampling and signal collecting were optimized, and the SNR of aroma fingerprint was improved. Different grading tea samples were measured with cross circulation array, with which to improve the suitability of the sensors of electronic nose and also eliminate the system error. After the samples were divided into calibration and prediction set, the SIMCA algorithm was used to build the classification model of tea grading. The results showed that the estimation accuracy reached to 95% for the model and proved that the intelligential classification method is applicable for Xihu-Longjing tea.