Unbounded data structures, advanced functions and data types, and/or different forms of communication are often needed to model large and complex probabilistic real-time systems such as wireless sensor network algorithms. Furthermore, it is natural to model distributed probabilistic real-time systems in an object-oriented style, including using subclass inheritance and dynamic object and message creation and deletion. To support the above features, we introduce probabilistic real-time rewrite theories (PRTRTs), that extend both real-time rewrite theories and probabilistic rewrite theories, as a rewriting-logic-based formalism for probabilistic real-time systems. We then show that PRTRTs can be seen as a unifying model in which a range of other models for probabilistic real-time systems---including probabilistic timed automata, stochastic automata, deterministic and stochastic Petri nets, as well as two probabilistic timed transition system models with underspecified probability distributions---can naturally be represented. We also provide semantics-preserving mappings from these models into PRTRTs, and prove their correctness. Finally, we show how the OGDC state-of-the-art algorithm for wireless sensor network algorithm can be specified in our formalism.