The Human Use of Music Information Retrieval Systems (HUMIRS) Project is a multi-year, multi-disciplinary research endeavour designed to address a major shortcoming in current Music Information Retrieval (MIR)research. Notwithstanding the many technological innovations being realized by the MIR community, there still is an almost complete absence of any empirical or theoretical information on how real-world users will be interacting with the wide-variety of MIR systems currently under development. For example, many research teams are developing Query-by-Humming (QBH) systems wherein users are supposed to construct their search statements by humming or whistling their music queries. Although there is a certain intuitive impetus for this kind of interface modality there is, in fact, no data showing that this approach meets any particular real-world user need. In Phase I of the HUMIRS project it is our goal to develop a taxonomic framework within which we can rigorously analyze the wide-ranging classes of real-world users, their information requests, and the modes of interaction they require. This Phase I framework will allow the construction of formal user models (Phase II). The formal user models will in turn provide a means by which developers can evaluate the real-world effectiveness of their MIR systems (Phase III).