Abstract: An important challenge regarding peer's trust valuation in peer-to-peer (P2P) networks is how to cope with such issues as dishonest feedbacks from malicious peers, collusions and complex strategic frauds. However, these issues cannot be effectively addressed by the existing solutions. Thus, a reputation-based distributed trust management model for P2P networks, named RATrust, is proposed to quantify and evaluate the trustworthiness of each peer, and to combat these attacks from the malicious peers in P2P networks. In RATrust, metric of time zone is used to describe the time property of the transaction experience and the recommendation. Three other metrics, such as peer trust value, short trust value and long trust value based on the metric of time zone, are applied to express accurately the final trust level of peers. Moreover, trust deviation value and trust abuse value are introduced to depict the above several trust metrics. Additionally, a reputation information storage mechanism is provided and discussed in this paper. Theoretical analysis and simulation experiments demonstrate that RATrust has advantages in combating various dynamic malicious behaviors such as collusive malicious behaviors, dishonest feedbacks and strategic malicious attacks, over the current trust model, and shows more effectiveness and stronger dynamic adaptability.