Telephony over IP has undergone a large-scale deployment thanks to the development of high-speed broadband access and the standardization of signaling protocols. A particular attention is currently given to P2PSIP networks which are exposed to many security threats. The RELOAD protocol defines a peer-to-peer signaling overlay designed to support these networks. It introduces a security framework based on certification mechanisms, but P2PSIP networks are still exposed to residual attacks, such as refusals of service. We propose in this work to address these residual attacks by integrating into the RELOAD architecture a dedicated trust model coupled with prevention countermeasures. We mathematically defines this trust-based strategy, and describe the considered prevention mechanisms implemented by safeguards and watchmen. We quantify the benefits and limits of our solution through an extensive set of experiments.