Vehicle logo classification is one of the methods of vehicle recognition. Vehicle logo classification is widely used in the safety protection of government buildings, campus, roads and other places. Aiming at the vehicle logo classification, a Pin-SVM based on large margin distribution (LMD-Pin-SVM) is proposed. Firstly, the margin distribution is expressed with margin mean and margin variance. Then introducing the optimal margin distribution into Pin-SVM and the linear LMD-Pin-SVM is built. In addition, the linear LMD-Pin-SVM is extended to the nonlinear case with kernel trick. And the nonlinear LMD-Pin-SVM is applied to vehicle recognition. All experiments show that the proposed method is superior to the state-of-the-art methods in generalization performance. Our LMD-Pin-SVM can effectively suppress the adverse effects of noise on the classification model and improve the classification accuracy of different vehicle logo images.