Over the past 20 years, many libraries have devoted significant resources to digitizing their special collections. Relatively few resources have been devoted to designing effective and easy-to-use online delivery platforms for these collections. Through an online questionnaire and a series of semi-structured interviews with University of Oxford humanities researchers, this dissertation investigates how academics discover digitized special collections, what resources they make use of, and what metadata and image functionality is most important to them. Key points include the importance of centralized shelfmark indices, the ineffectiveness of search engines in crawling digital library platforms, and the variance in user needs depending on discipline, career stage and geographic location.