Cost effectiveness of the QUERY program for searching the Educational Resources Information Center (ERIC) data base has been an important issue at Auburn University. At least two broad categories of costs are associated with information retrieval from a data base such as ERIC: fixed costs or overhead and data base associated costs. The concern at Auburn reflected in this paper, has been particularly with three data base associated costs: preparation of records before searching, selection of a software search system, and implementation of search strategy. Computational algorithms associated with costing particular jobs should have terms applicable to each of these variables. The first two are rather straightforward to implement, but the third, because it is a human variable, is difficult to access. The most powerful approaches to cost reduction seem to be associated with partitioning of the data base into components specific to the needs of the user population. (Author/SJ)