This paper proposes a novel synchronization algorithm for wireless sensor networks (WSNs), the Maximum Time Synchronization (MTS), which is based on maximum value consensus approach. The main idea is to maximize the local information to achieve a global synchronization. Compared with the existing consensus-based synchronization protocols, the main advantages of our protocol include: i) a faster convergence speed such that the synchronization can be completed in a finite time; ii) simultaneous compensation for both the skew and the offset. We provide a rigorous proof of convergence to global synchronization and also give the upper bound of the convergence time. Moreover, the protocol is completely distributed, asynchronous, and robust against packet losses, nodes failure and the addition of new nodes. Some numerical examples are presented to demonstrate the efficiency of our protocol.