A document retrieval system should rank documents in order of their usefulness or satisfaction to the users. This principle was first explicated in the classic paper by Maron and Kuhns (1). Additional considerations concerning document ranking have been suggested by other researchers (2,3). Particular attention will be given here to the ranking algorithm appropriate for those presenting the same request, but having different information needs. The research on which this report is based identifies limitations associated with sequencing rules that use a probability ranking technique (4). Three basic and somewhat interdependent limitations will be discussed.