In this paper, we propose an efficient peer-to-peer information retrieval system PeerSearch that supports state-of-the-art content and semantic searches. PeerSearch avoids the scalability problem of existing systems that employ centralized indexing, index flooding, or query flooding. It also avoids the non- determinism that exhibited by heuristic-based approaches. PeerSearch achieves both efficiency and determinism through an elegant combination of index placement and query routing. Given a query, PeerSearch only needs to search a small number of nodes to identify matching documents. 6 Pages