||Object-Oriented Databases (OODBs) have advanced very quickly in the last decade. They have been employed in more and more applications. The Web and data warehouse are just two of the many examples. We can expect that a lot of computational resources will be saved if we can improve the efficiency of query processing on OODBs. This thesis focuses on the development of a new index organization to index a data graph, which represents the underlying objects and object relationships corresponding to a path in the object schema, so that the OODB can be searched more efficiently. The key idea of our index organization, called Partitioned Flexible Object Range Tree (PFORT), is to map the data graph into some spatial objects on a 2-D plane, which can then be indexed using some spatial data structures. This approach is quite different from the traditional index methods for OODBs. The advantage of PFORT is its search efficiency and its ability to support other functions such as range queries on attributes. The performance evaluation shows that compared with other traditional index methods PFORT has a much better retrieval performance at a reasonable storage overhead especially when the fanout and the number of objects in each class are high.