This paper took soy sauce as the research object to study the application of the electronic tongue technology on soy sauce. Principal component analysis（PCA）， discriminant factor analysis（DFA） and BP neural network （BPNN） were used to identify different brands of soy sauce. Partial least squares method（PLS） was used to establish a quantitative model of the taste components such as amino acid nitrogen， total acid， total sugar， salt， bitter amino acids， umami amino acids and the output value of electronic tongue to achieve the purpose of predicting the content of taste components. The optimal conditions for the electronic tongue to identify soy sauce were as follow： the dilution multiple was 30 times， sourness， bitterness， umami， saltiness， sweetness and richness were chosen as the taste indexes for evaluation， and washing time before full sensing was 6 s. The results showed that the electronic tongue technique had the ability to classify soy sauce in different brands. The contribution rate of the first two principal components of PCA and DFA were 83.8% and 98.1%， respectively， and the correct discrimination rate of the fishers’ function was 99.3%. BPNN had the ability to predict unknown samples accurately and the correct discrimination rate reached 100%. It could be seen that the PLS model showed excellent ability to predict the contents of physicochemical indexes accurately by the electronic tongue.