A new approach of classification of engineering rock based on the support vector machine is presented. The nonlinear mapping between rock classification and its conditions are learned from the finite samples and used to classify the rock. The precise results are acquired by using this approach.