In order to improve cutting tool condition pattern recognition, the face milling cutter is taken as an object and the feature of the milling force signal is analyzed using an ant colony algorithm. A method for feature selection that can be used in pattern recognition of tool wear is proposed. The method transforms the feature selection into a search for the best routes in ant colony algorithm, and the formula for this selection route is given. The fisher criterion is adopted as heuristic information. At the same time, the optimal feature subset of the current iteration cycle is put onto a BP neural network for Pattern Recognition. The precision of classification is obtained and used in the policy of pheromone update. Moreover, the method of parameter selection on the colony algorithm is improved. The experimental results show that this scheme can efficiently obtain the optimal feature subset. The accuracy is significantly higher than that obtained without feature selection.