With the application of the Spatial Information Grid (SIG),the spatial information managed by SIG being more andmore abundant. The abundantspatial information reguires the betterapplication ofdistributed spatial information query acrossSIG. However, the remote spatial join queries are always the bottleneck in the distributed spatial information query. Based on this observation, in this paper,the spatial join queries are optmi ized by taking full advantage of the grid computing resources according to the characteristics of spatial information.At first, the software architecture for distributed spatial query is designed based on the different grid services. The distributed spatial data query software architecture is composed of three different kinds of grid services, namely:Distributed SpatialDataQuery Grid Service (DSDQGS), SpatialData Grid Service (SDGS) and Remote Spatial Join Query Grid Service (RSJQGS), These three kinds ofgrid services cooperate to mi plement the optmi ization and execution of the distributed spatial data query.In the architecture,the grid computing resources are utilized by the remote spatial join queries execution grid services.Secondly, the partitioned parallel spatial join queries are mi plemented by theKd-Tree spatial partition scheme. In the scheme, an original spatial query is rewrite into several sub-spatial queries bounded by sub regions of the Kd-Tree nodes, which can be run concurrently; therefore, the performance of the remote spatial join queries is mi proved. The cost model for the partitioned parallel spatial join queries is also presented in paper.The costof the remote spatial join query involves two parts: the computing costof the join operation and communication costof the spatial data.Thirdly, the optmi ization algorithm for the query planed to generate the remote spatial join queries is designed according to the costmode.l The remote spatial join query plan prescribes the way the spatial join query execution,including the scheme forpartitioned parallel spatial join query, SDGSs participated in the join query, and assignments of\nthe partitioned parallel spatial join query tasks to RSJQGSs. The cost isbenchmark for the remote spatial join query plan.The parameters utilized in the optmi ization algorithm are managed as properties of the WSRF. The full optmi ization algorithm is finishedwhen all identified spatial join operators are processed.At last, the future research directions for the optmi ization of spatial distributed query on SIG are discussed.