Current P2P data sharing systems process queries issued by peers with data granularity of files. However,these methods cannot implement advanced query processing on finer granularity datasets, such as relational data, XML etc. In this paper, we propose a P2P view-based query processing framework on relational datasets: (1)peers issued the same query (or subquery) build views through collaborative scheme; (2)peers then answer queries directly from these views. It is proved that the view-based query processing in P2P environment can greatly reduce the bandwidth consumption on eacb. peer, and greatly improve the response time of queries.