In order to enhance the correct rate of identification result of six kinds of vinegars, a kernel Fisher discriminant analysis (KFDA) method is introduced. And a measuring method of matrix similarity based on distance discrimination was presented to define the radial basis function (RBF) characteristic parameter, where RBF was selected as the kernel function. The measuring method is that, firstly, an ideal Gram matrix is defined, and the actual kernel Gram matrix is calculated by RBF; secondly, Euclidean distance can be employed to measure the degree of approximation between the actual kernel Gram matrix and the ideal Gram matrix; finally, the optimal kernel parameter can be obtained by extremal solution of the distance. When two kinds of feature vectors, whose were integral value and average value in relative steady state, were extracted from the E nose signals of vinegar samples, the corresponding characteristic parameters were 5.7700 (integral value) and 5.3878 (average value in relative steady state). Comparing and analyzing the results of Fisher discriminant analysis (FDA) and KFDA, their identification correct rates were respectively from 93.3% and 90.6% (FDA) up to 98.3% and 98.3% (KFDA). This indicates that the suitable KFDA method can effectively improve the identification results of the six kinds of vinegar samples.