Traditional image subcategory classification methods combined multiple features into a feature vector. Such methods neglect distinct roles of diverse features on discriminating image subcategories. In this paper, the Dempster-Shafer evidence theory is applied to fuse different features in image subcategory classification. It considers the different contribution of each feature to image classification with limited samples. The experimental results on car subcategory classification show that our proposed method outperforms the k nearest neighbor algorithm in terms of classification accuracy.