In advanced applications like geographic information system (GIS), multi-attribute indexing structure is required. To support the various multi-attribute applications such as business, GIS, or multimedia, the components of conventional DBMS should be extended to effectively handle multi-attribute data. In this paper, we point out the limit of conventional multi-attribute indexing structures: poor storage utilization and approximated regions in point data, spatial overlaps and spatial duplication in spatial data. To overcome the limit, we propose a new flexible indexing structure which is applicable to the various multi-attribute applications.