Abstract. This paper presents a unifying framework to model casebased reasoning recommender systems (CBR-RSs). CBR-RSs have complex architectures and specialize the CBR problem solving methodology in a number of ways. The goal of the proposed framework is to illustrate both the common features of the various CBR-RSs as well as the points were these systems take different solutions. The proposed framework was derived by the analysis of some systems and techniques comprising nine different recommendation functionalities. The ultimate goal of the this framework is to ease the evaluation and the comparison of case-based reasoning recommender systems and to provide a tool to identify open areas for further research.