Advances in web technology have given rise to new information retrieval applications. In this paper, we present a model for geographical region search and call this class of query similar region query. Given a spatial map and a query region, a similar region search aims to find the top-k most similar regions to the query region on the spatial map. We design a quadtree based algorithm to access the spatial map at different resolution levels. The proposed search technique utilizes a filter-and-refine manner to prune regions that are not likely to be part of the top-k results, and refine the remaining regions. Experimental study based on a real world dataset verifies the effectiveness of the proposed region similarity measure and the efficiency of the algorithm.