The purpose of this paper is to present a new method for fuzzy entropy measure. The proposed fuzzy entropy is modified from Shannon's entropy. The definition of fuzzy entropy based on the entropy of fuzzy sets is employed to deal with the fuzzy events. The probabilities of fuzzy entropy are measured via the proportion of the membership function values of occurred elements. For classification problem, it has the ability to reflect the actual distribution of patterns. A practical example is given to illustrated the accuracy of the proposed method.