Abstract: OBJECTIVE To study the taste of goji from different producing regions and establish a taste discrimination model. METHODS Samples of goji were collected from Ningxia， Inner Mongolia， Qinghai and Gansu. Five sensory attributes (sour， salty， fresh， sweet， and bitter) were evaluated by artificial tasting， and the response values of 7 sensors (AHS， ANS， PKS， CTS， NMS， CPS， SCS) representing the taste of pieces were detected by electronic tongue. The correlation between sensory evaluation and response values of the electronic tongue were analyzed. Principal component analysis (PCA)， partial least square discriminant analysis (PLS-DA) and orthogonal partial least square discriminant analysis (OPLS-DA) were used to process and analyze the response values of the four producing areas， and a Fisher discriminant model was established to distinguish four producing areas.RESULTS Correlation analysis showed that the sensory evaluation of sour， salty， fresh， sweet， bitter and AHS， CTS， NMS， ANS， SCS sensors had different degrees of correlation (r＞0.6， P＜0.01). This indicated that the electronic tongue technology could be used as an alternative of artificial taste; Goji from Gansu and Inner Mongolia could be distinguished by PCA; Goji from four regions could be distinguished by PLS-DA and OPLS-DA of goji from Qinghai and Ningxia; The Fisher discriminant model for goji which from different producing regions was established， and it could effectively discriminate them with a recognition rate of 100%. CONCLUSION Electronic tongue technology can replace artificial taste to evaluate the sour， bitter， salty， sweet and fresh of goji， and provide a fast and convenient technical means for the differentiation of goji from different regions with multivariate statistical analysis.