Abstract DECADE is a prototype knowledge-based system for catalyst selection. From a specified reaction it attempts to propose a set of materials with high probability of being good catalysts for the input reaction and the conditions at which the proposed catalysts should operate. While the first paper on DECADE (Bañares-Alcántara et al., (1987) concentrated on the organization and control aspects of DECADE from the expert system's point of view, this paper presents in more detail the organization of the program and, in particular, the selection strategy from the chemistry point of view. Specifically, there are three levels of responding to a request for catalyst selection: 1. 1. Responding with catalysts that are known to work from published experimental results and that are in the DECADE knowledge base. 2. 2. Classifying the reaction from generic to specific classes and recommending catalysts known to catalyze reactions in all of these levels of classification. 3. 3. Determining both required and unwanted surface steps (adsorption, desorption, dissociation, addition) for the reaction, and selecting materials known to catalyze the required steps and to suppress the unwanted ones. The most interesting results are produced at the deeper level of abstraction (which relies only on medium to deep level knowledge). Catalysts are proposed at this level using a generate and test procedure with a priori and dynamically-generated constraints. Explanation of the results is available for any material that was taken into consideration (whether recommended or not). In view of the limited size of the knowledge base, the results and explanations for this level of abstraction are satisfactory. The operation of DECADE is achieved through the interaction of several modules written in different languages and communicated by means of a blackboard system. The size of DECADE is roughly 300 production rules (in OPS5, Forgy, 1981), 325 frames (in SRL, Wright and Fox, 1983) and 200 functions (in FranzLisp, Foderaro, 1981).