This paper discusses the representation of a variety of role-based access control (RBAC) security models in which users and permissions may be assigned to roles for restricted periods of time. These security models are formulated as logic programs which specify the security information which protects data, and from which a user’s permission to perform operations on data items may be determined by theorem-proving. The representation and verification of integrity constraints on these logic programs is described, and practical issues are considered together with the technical results which apply to the approach.