Product configuration systems are an emerging software technology that supports companies in deploying mass customization strategies. In this paper, we describe a CLP-based reasoning engine that we developed for a commercial configuration system. We first illustrate the advantages of the CLP approach to product configuration over other ones. Then, we describe the actual encoding of the considered product configuration problem as a constraint satisfaction problem. We devote a special attention to the key issues of constraint propagation and optimization as well as to the relevant process of assignment revision. A comparison with existing systems for product configuration concludes the paper.