Associative file structures are potentially valuable for many database and artificial intelligence applications but very little information is available to database designers trying to choose an appropriate file structure for a particular problem this paper describes an experiment comparing the retrieval performance of k-d trees quad-trees an flat files as measured by CPU time, wall clock time and I/O operations five types of queries are used: exact match partial match range search, nearest neighbor, and best match the database used in this study is astatic medical database of half a million characters with the patient information removed results suggest that there is no one best type of file structure for all types of associative queries; quad trees dominated with some query classes, k-d trees with others.