Abstract: The states of metal cutting, such as the chip shapes （C-shape, coiled, tangled chips）, the existance or non-existance of BUE （Build-Up Edge）, the vibration of processing, are muddy and fuzzy phenomena. In this paper, the fuzzy clustering method of fuzzy equivalent matrix is adopted for protype classification. Therefore, the centers of class are found. Then, the distances between the centers of class and the sample to be recognized are examined so as to find out which class the sample belongs to. The feature selection in the pattern recognition, viz. filtering the unrelated feature and the derived feature, is discussed. Also, the solution of the determination of the feature weights is obtained by use of the method called"the function of the feature representing the degree of homogeneity", in result a scientific basis for determining weights of features is provided. The paper shows successful application of the fuzzy mathematics to the pattern recognition of the metal cutting states.