In order to quickly identify different taste qualities of beef juices and divide the quality grade of beef individuals, a taste sensor array was built by the method according Byrd Luo taste theory, containing twelve working electrodes and one reference electrode. The array was applied to classify the quality grade of beef juices, and 30 groups of beef juices were distinguished. The method of clustering analysis based on Euclidean distance was used to classify the taste quality of beef, which showed that the taste characteristics of the same grade of beef juice samples were very similar and could be aggregated into one group. And the accuracy of sensor array to detect the beef taste quality was assessed by sensory evaluation. The sensor array for identification of beef juices was optimized for the key technical issues of automated quality evaluation of beef juices. The inherent relationship among the response signals of sensors was analyzed by the factor analysis of variance. Six working electrodes (S1, S2, S5, S7, P2 and P3) were selected to compose beef taste sensor array with the reference electrode, and 30 groups of beef juices were identified. The results showed that the identification accuracy rate for beef juices by the optimized sensor array was 93.33%, which was higher than 80.00% by the non-optimized one.