Abstract: For efficient evaluating the reputation of the target peers in Peer-to-Peer environments, an encourage-and-punish-based computational reputation model is proposed. Th proposed model considers not only the direct transactions, but also the recommending data of other peers, which includes their direct transactions with target peers and their querying results before their transactions. By different weights on these data and the measure of relative encouragement and punishment, the integrated reputation of peers may be computed. The experimental results show that the present model can not only hold back malicious peers, but also ensure higher transaction rate.