A group-based semantic search method ?GBSS is introduced to improve the efficiency and scalability of search in P2P systems. GBSS introduces a hybrid architecture in which the storage and search of raw data is based on DHT network, but the storage and search of meta-data are based on unstructured P2P network. With regard to search algorithm, GBSS introduces a new file grouping method requiring no users? descriptions and constructs friend group relations between nodes based on interest similarity between file groups. The forwarding of users? requests makes use of friend group relations. Simulation tests show that compared with traditional flooding algorithms, GBSS is both efficient and scalable.