We present a knowledge-based system for reactor selection. First the system makes an analysis of the reaction network and derives a set of strategic notions that define a profile for the most desirable reactor type for the chemical conversion process. Then it uses explicit knowledge about technical reactors to select the reactor that is best suited. This knowledge encompasses technical and practical aspects of reactors. The system is specially geared to incorporate the practical knowledge of the engineers in industry. The fact that reactor selection is a creative task is honoured by the system. The selection problems do not proceed along a fixed pattern, but are allowed to develop in their own way. The system supports a 'what-if' type of reasoning such that consequences of different assumptions can be explored. We demonstrate its use in a detailed example. A prototype system has been implemented in prolog.