It is hard to optimize query latency, query hit, and query cost at the same time for the resource location of unstructured peer-to-peer network. For this problem, this paper presents a probabilistic routing algorithm called DCBF (data copying and Bloom Filter), which is based on data copying and a Bloom Filter technique. DCBF makes a few copies of each shared resource and places each copy on a random selected node, based on a directed random network. Each node forwards membership information to neighboring nodes with distributed declining Bloom Filters. Analysis and experimental results show that DCBF can make the most of the nodes, use the membership information of resource objects by making only a few copies, and forward membership information with distributed declining Bloom Filter to achieve high query hits with low cost and low latency.