The support vector machine(SVM) based on structural risk minimization is more and more widely used to solve the problems of small sample, nonlinear, high dimensional and local minimization attributes because of its good generalization. But the performance of SVM is influenced by the model parameters very much. At present there is not a unified method of model selection, which makes it troublesome in the application of SVM. The paper compares the joint influences on SVM imposed by the radial basis kernel function and the penalty factor and by the scaling kernel function and the penalty factor, which is of some referring value to the selection of the model parameters of SVM.