Aimed at the problem that the existed spatial join algorithms can not solve the spatial join query within the constrained range, the constrained spatial join query is proposed which finds all the pairs of objects satisfying some spatial predicate within the given range. The directed solving methods and algorithms based on R-tree are given. Based on good property of QR-tree, a constrained spatial join algorithm is proposed which avoids larger storage cost of quadtree and overcomes the drawbacks of R-tree node overlapping. Thus the algorithm implements the constrained spatial join query on many small R-tree and the problem of expensive spatial join overhead is solved. The cost analysis for the proposed algorithms is given. Experiments show the algorithm has high efficiency.