In this paper, the emerging P2P computing is first briefly introduced, including its distinct features, potential merits and applications, and the problems from which the existing P2P-based data sharing systems are suffering are further point out. To address these problems, the concept of P2P-based information retrieval is proposed, which can not only exploit the potential merits of P2P to overcome the problems of traditional information retrieval systems (e.g., lacking of scalability), but also achieve fully semantic retrieval and sharing in the context of P2P systems. Based on the ideology, PeerIS, a P2P-based information retrieval system is developed. Then, the architecture of PeerIS and its peers’ components are presented. The key issues of implementation are described, including communication, semantics-based self-reconfiguration, query processing and self-adaptive routing mechanisms, are also described. Finally, an experimental study is used to verify the advantages of PeerIS.