In order to handle effciently spatial data, such as geophysical data, geologicaldata, and computer-aided desigin data, a database system needs an index mech-anism that will help to retrieve data items more quickly according to their spatiallocations. Hovever, traditional index methods are not well suited to data objectslocated in multi-dimensional space. In this paper the authors discuss a dynamicindex structure called the R-tree which meets this need, and discuss the algorithmsof searching, adding, deleting and application of R-tree for fault data handling.