Semantic Web Services enable the dynamic discovery of services based on a formal, explicit specification of the requester needs. The actual Web Services that will be used to satisfy the requester’s goal are selected at run-time and, therefore, they are not known beforehand. As a consequence, determining whether the selected services can be trusted becomes an essential issue. In this paper, we propose the use of the Peertrust language to decide if trust can be established between the requester and the service provider. We add modelling elements to the Web Service Modeling Ontology (WSMO) in order to include trust information in the description of Semantic Web Services. In this scenario, we discuss different registry architectures and their implications for the matchmaking process. In addition, we present a matching algorithm for the trust policies introduced.