— There have been many time synchronization protocols proposed for sensor networks. However, the issues related with securing such protocols have not received adequate amount of research attention. If malicious entities can manipulate time synchronization, failures of many functionalities of the sensor networks would occur. In this paper, we identify various attacks against time synchronization and then develop a detection and self-healing scheme to defeat those attacks. The proposed scheme has three phases: (1) abnormality detection performed by individual sensors, (2) trustbased malicious node detection performed by the base station, and (3) self-healing through changing topology of the synchronization tree. Simulations are performed to demonstrate the effectiveness of the proposed scheme as well as the implementation overhead.