Wireless sensor networks pose numerous fundamental coordination problems. For example, in a number of application domains including homeland security, environmental monitoring, and surveillance for military operations, a network's ability to efficiently manage power consumption is extremely critical because direct user intervention after initial deployment is severely limited. In these settings, limited battery life gives rise to the basic coordination problem of maintaining coverage while maximizing the network's lifetime. In this paper, we propose a distributed scheme for efficient power management in sensor networks that is guaranteed to identify suboptimal topologies in an online fashion. Our scheme is based on a general (game-theoretic) mathematical structure that induces a natural mapping between the informational layer and the physical layer. We provide sufficient conditions for the convergence of the algorithm to a pure Nash equilibrium and characterize the performance of the algorithm in terms of coverage. We also present encouraging performance results on a MicaZ testbed as well as on large-scale topologies (obtained via simulation).