In this paper we propose the use of process algebras as powerful frameworks for model-based diagnosis. In fact, they provide machinery and tools for building component-oriented models, for characterizing and computing diagnoses, and for analyzing properties such as the diagnosability of the system under investigation.