This paper proposes a query in wireless sensor networks: Peak Region Query (PRQ). Given the shape and size of the query region, i.e., a disk region with radius R, peak region query finds out a region with this shape in the network field, in which the aggregation value of the data of the sensors can be maximized. This paper first gives the definition of PRQ, and then proposes a centralized algorithm for the problem. Because the sensors have limited energy, a distributed approach EXQ (an algorithm for extreme value query processing) is proposed, which not only reduces the energy cost but also balances the workload of the sensors, so as to prolong the lifetime of the network. The basic idea is to divide the network field into overlapped sub-regions, compute a local result for each sub-region and aggregate these results to obtain the query answer. The paper compares the energy efficiency and load balance between EXQ and the centralized approach analytically and experimentally.