Historically, bibliographical classification and indexing systems have concentrated on textually-based materials. More recently, archives and libraries have realized a pressing need to provide access to visual documents. An institution like the Ohio Historical Society must consider many different tools of access to achieve this goal. The growth of the Internet makes it an attractive conduit for providing access via an electronic index of bibliographical surrogates of the collection. The heart of the index is the bibliographical record. If the search terms selected by the researcher do not match the terms selected by the indexer, the index is a failure. This study addresses methods of anticipating the language and needs of an index's future users by charting and analyzing past information requests. It also examines the different sources available for generating the controlled vocabulary and bibliographic record format for image collections. Finally, it describes the creation and testing of an electronic index by four Ohio Historical Society staff researchers. Findings include: the majority of reference requests are very specific; the subject headings of the "Library of Congress Thesaurus for Graphic Materials" are the most appropriate for this collection; and the best model for the bibliographic record is the Library of Congress record. Several modifications of these records proved appropriate in the areas of field names and the level of specificity of subject headings. Appendices include reference and information-only requests from July 1995-May 1997, sample records, and a bibliography of Civil War Ohio reference books. (Contains 28 references.) (Author/SWC)