AbstractThe challenges of design of programming systems, file organization and manipulation, and user‐oriented query languages have held the spotlight in the development of data processing support to systems for retrieval of scientific literature. At BIOSIS, development of such a system from the pragmatic viewpoint of providing viable self‐supporting retrieval service from a data base of more than two million references has resulted in focusing attention on supporting the formulation of the information need into an effective computer search specification. Developed under CPS and now implemented under CICS, the program claims no novelty of structure or technique, but constitutes an innovation in application and goal with primary emphasis on accommodating the behavior of a user not previously knowledgeable about the structure, indexing language, and detailed content of a machine based references f