Five grades of Xihulongjing tea (grade: AAA, AA, A, B and C, from the same region and processed with the same processing method) were discriminated using -Astree II electronic tongue (e-tongue) coupled with pattern recognition methods including principal component analysis (PCA), canonical discriminant analysis (CDA) and back-propagation neural networks (BPNN). Results of PCA and CDA showed that the grades of the samples were discriminated with the exception of a few overlap samples between grade AA and grade A. The discrimination of accuracy of the training sample set and the predicted sample set was 95.7 and 97.5%, respectively, by the analysis of BPNN. 92.9% of all the crossvalidated training sample set and 100% of the predicted sample set were exactly grouped by CDA. The sensory evaluation of the samples was consistent with the evaluation based on the e-tongue. The results show that the e-tongue is a potential tool to identify the tea quality.