Aiming at the wide application of support vector machine (SVM) model in classification, a new personal credit evaluation model based on SVM is proposed.The voting matrix is constructed by combining four kinds of kernels, namely, histogram cross kernels, thermonuclear feature kernels, Jacquard distance kernels and cosine generalized distance kernels.Through the actual data experiment, we get good classification results, and prove that the support vector machine adaptive combination kernel weighting model has good performance in the credit scoring system.Therefore, this personal credit evaluation model based on support vector machine can really help banks or lenders make correct decisions.