Abstract: Aim To improve the query processing of heterogeneous distributed database. Methods An approach were explored to get join and semijoin sequences which mini- mize the amount of data transmission required for heterogeneous distributed query pro- cessing. The algorithm selected the most hopeful node as next expended node in a search space through using a lot of heuristic rules, which minimized the amount of node searched. Results An important concept was proposed which occur with the use of join operations as reducers in query processing, namely, useful semijoin. Some nonprofitable semijoin, though not profitable in themselves, become profitable owing to the use of subsequent join operations. Conclusion This method can improve the que- ry processing distributed database, specially on WAN.