Hybrid rendering of volume and polygonal model is an interesting feature of visualization systems, since it helps users to better understand the relationships between internal structures of the volume and fitted surfaces as well as external surfaces. Most of the existing bibliography focuses at the problem of correctly integrating in depth both types of information. The rendering method proposed in this paper is built on these previous results. It is aimed at solving a different problem: how to efficiently access to selected information of a hybrid model. We propose to construct a decision tree (the Rendering Decision Tree), which together with an auxiliary run-length representation of the model avoids visiting unselected surfaces and internal regions during a traversal of the model.