: For the characteristics of malfunction diagnose system a model to classify fault printing based on support vector machines is discussed. The printing malfunctions have many classes. To use multi-layer classification machine based on BSVM (binary-class support vector machine) to set a BSVM for every class. It executes classification from the top down for every layer and every class. After classified by SVM, it uses C4.5 algorithm for further diagnosis. The hybrid method makes full use of the SVM efficiency in multidimensional character space. Simultaneously, it also brings the accuracy of C4.5 algorithm into full play. That is suitable for the complicated print fault. Experimental results demonstrate the method has a good efficiency for adjusting faults.