The effective classification of image contents allows us to adopt those strategies that can best satisfy the increasing demand for quality, speed and ease of use in imaging applications. We present here the results of our experimentation using Cart trees for the classification of images indexed by low-level pictorial features, such as color, texture, and shape. Our study addressed the high-level problem of distinguishing photographs, graphics and texts for an application in the context of cross-media color reproduction. The results obtained to date are very good in terms of accuracy, and also demonstrate the strength of the approach in providing information that can be used to reduce the dimensions of the feature space.