Since the 1980s, two approaches have been developed for analyzing security protocols. One of the approaches relies on a computational model that considers issues of complexity and probability. This approach captures a strong notion of security, guaranteed against all probabilistic polynomial-time attacks. The other approach relies on a symbolic model of protocol executions in which cryptographic primitives are treated as black boxes. Since the seminal work of Dolev and Yao, it has been realized that this latter approach enables significantly simpler and often automated proofs. However, the guarantees that it offers have been quite unclear. For more than twenty years the two approaches have coexisted but evolved mostly independently. Recently, significant research efforts attempt to develop paradigms for cryptographic systems analysis that combines the best of both worlds. There are two broad directions that have been followed. {\em Computational soundness} aims to establish sufficient conditions under which results obtained using symbolic models imply security under computational models. The {\em direct approach} aims to apply the principles and the techniques developed in the context of symbolic models directly to computational ones. In this paper we survey existing results along both of these directions. Our goal is to provide a rather complete summary that could act as a quick reference for researchers who want to contribute to the field, want to make use of existing results, or just want to get a better picture of what results already exist.