We propose SARP, a provably incentive-compatible and energy-efficient routing protocol for ad hoc networks. The key distinguishing factor when compared to previous proposals is the practicality and (relative) scalability of SARP due to a small communication complexity (analytically and practically), which is achieved by computing routing paths in a novel manner that borrows techniques from geometric routing. We also propose a solution on how to deliver payments (rather than only computing them). We support our claim that SARP works in realistic settings through comparative simulation results obtained from an implementation of SARP in GloMoSim.