Abstract. We provide a framework for modeling and analyzing both qualitative and quantitative aspects of mobile ad hoc network (MANET) protocols above the data-link layer. We extend Restricted Broadcast Process Theory [11,9]: delay functions are assigned to actions, while the semantics captures the interplay of a MANET protocol with stochastic behavior of the data-link and physical layer, and the dynamic topology. A continuous-time Markov chain is derived from our semantic model by resolving non-determinism, using the notion of weak Markovian network bisimilarity. The framework is applied to a leader election algorithm.