We develop a formal framework for the analysis of security protocols in wireless networks. The framework captures characteristics necessary to reason about neighbor discovery protocols, such as the neighbor relation, device location, and message propagation time. We use this framework to establish general results about the possibility of neighbor discovery. In particular, we show that time-based protocols cannot in general provide secure neighbor discovery. Given this insight, we also use the framework to prove the security of four concrete neighbor discovery protocols, including two novel time-and-location-based protocols. We mechanize the model and some proofs in the theorem prover Isabelle.