The simplicity and robustness of unstructured P2P system make it a preferable architecture for constructing real large scale file sharing system. Most of the existing paradigms require more overhead. The trace data analysis shows that the workloads among peers are correlated. The bigger the intersection of workloads, the higher the probability that they share other files is. By exploiting such principle, we propose IAC, an interest-aware resource advertisement caching paradigm for unstructured p2p system. Each peer advertises its resource list. If a peer is interested in the resource advertisement received, it then caches the advertisement. Through local cache search, usually the peer gets more than 50% success rate. If local cache search is failed, the random walk-based search is used. The simulations show that as to the same hit rate, the maintenance and search overheads are low and the search delay is very low.