In this paper a self-stabilizing program for solving a pursuer-evader problem in sensor networks is presented. The program can be tuned for tracking speed or energy efficiency. In the program, sensor motes close to the evader dynamically maintain a "tracking" tree of depth R that is always rooted at the evader. The pursuer, on the other hand, searches the sensor network until it reaches the tracking tree, and then follows the tree to its root in order to catch the evader.