Remote monitoring of physiological data of patient is emerging as technology called Wireless Body Area Networks (WBAN). As WBAN devices are operated in hostile environments, providing security and privacy to patients are challenging tasks. Due to simple, low cost and resource constrained nature of the sensors of WBAN, adversary can easily compromise one or more nodes and make clones of compromised nodes to launch different insider attacks in the network. In this paper, we propose a clone detection and prevention strategy for WBAN by leveraging inconsistent collisions so that legitimate nodes of WBAN are alone allowed to participate in communication, while preventing the cloned nodes. Through simulation, we show that the proposed algorithm can detect cloning attack fairly fast and with required accuracy under various conditions.