Abstract: This paper presents a way which recognizes the Chinese characters in handwriting on basis of statistical decision making and sentence structure.The method of recognition can work because a Chinese character can be represented by a grammar tree.Each pattern of Chinese character can be described by generating a grammar tree through separating the roots of the character and the recognition of them.The procedure of the recognition is as follows:firstly,"gaps of structure" of a Chinese character are searched to separate the roots of the character,and then,the roots separated are clustered fuzzily,compared and recognized according to the theory of fuzzy mathematics,finally,the grammar tree is traveled symmetrically,the Chinese character is assembled,the Chinese dictionary in the computer is looked upon and the result of recognition is output.This method of recognition has been proved through the analogous experimentation on micro-computer Apple Ⅱ.