Abstract: The integration of remote sensing and GIS is the long-term goal of both remote sensing and GIS.Remote sensing is the important data source of GIS and the means of data updating.Additionally, GIS is the auxiliary information of remote sensing image processing and classification, which is used to automatically extract semantic information.In the history of the development of remote sensing there are three different kinds of integrations.One is the separated but parallel integration (different user interfaces, different tools and different databases).The second is seamless (the same user interface, but different tools and different databases).The third is total integration (the same user interface, tools and database).However, all the integration is aimed at the single image or the integrated management of two data in one workspace. With the rapid development of remote sensing and digital photogrammetry technologies, the amount of image data has dramatically increased, and has garnered more and more attention.For example, one basic idea of Digital Earth is to cover the surface with high-resolution remote sensing images, and by means of DEM covering the world, to establish the virtual landscape of the earth.On the other hand, the cybercity has gained a much higher profile.In the establishment of these models, the integrated management of image data, DEM and vector data are in great demand, not just for multi-sources data management of a single image or a small area, but for management covering a city, a province, a country and even the whole world.The commercial GIS software packages can handle the large spatial database with vector data format.However, most of them are not efficient enough to manage image database and DEM database.Therefore, it is necessary to develop a spatial database management system that can integrate various data. With regard to the development of contemporary computer technology, the hardware fully meets the needs of multi-sources integrated management.The external memory of computers is able to extend to tens of TBs, and even thousands of TBs.The current urgent problem is that the commercialized GIS software cannot meet the demand.Popular GIS software currently on the market has already developed the mature technology to manage vector database.However, it has not yet been adapted to manage image database, DEM database and the integrated three databases. This paper brings forward a vector, image and DEM integrated spatial data model on the basis of the object-oriented idea.It also discusses the implementation method of this integrated spatial database management system.The spatial database management system has been used in the construction of China Spatial Data Infrastructure as the base of a GIS software, GeoStar.It reveals that the system has high efficiency and feasibility to manage national, provincial and city spatial data of multi-scales and multi-sources.