Remote monitoring and gathering information are the main objectives behind deploying Wireless Sensor Networks (WSNs). Besides WSN issues due to communication and computation restricted resources (low energy, limited memory computational speed and bandwidth), securing sensor networks is one of the major challenges these networks have to face. In particular, the security of sensors localization is a fundamental building block for many applications such as efficient routing. In this paper, we introduce a new threat model that combines classical Wormhole attacks (i.e. an attacker receives packets at one location in the network, tunnels and replays them at another remote location using a powerful transceiver as an out of band channel) with false neighborhood topology information sent by the wormhole endpoints themselves or by some colluding compromised nodes. We show using intensive simulations how this clever attacker that would exploit the neighborhood topology information can easily defeat two representative secure localization schemes. We also present some possible countermeasures and the first corresponding results.