Abstract: There are some instinctive characteristics in unstructured P2P network, such as frequent reappearance of keywords and uneven resource sharing. To enhance performance of resource searching, a Response Tracing based Resource Searching Mechanism (RTRSM) is proposed, which traces the content and the latest source of responses to construct an informed search. RTRSM adaptively forwards the query to the suitable neighbor, according to the related degree of the storing information and the query itself. By constructing an informed search, RTRSM hopes to maintain high success rate, decrease searching cost and promote searching efficiency. Simulation results show that RTRSM is scalable. It can be extended to large scale unstructured P2P network with a low cost. Simulation results also show that RTRSM achieves good searching efficiency. RTRSM decreases 54.2% of average message counts and 56.4% of average hops by contrast to flooding, and decreases 28.9% of average message counts and 6.4% of average hops by contrast to SQR (Scalable Query Routing), while the success rate keeps up to 90%.