The Australian Federal Government's Office for Learning and Teaching (OLT) has built an important collection of learning and teaching resources for the higher education (HE) sector, a product of the many projects OLT and its precursors, including the Australian Learning and Teaching Council and the Carrick Institute, have funded over the past two decades. Although these resources are freely available on its website, the OLT considers them underutilised. Hence it has commissioned a project to reorganise the collection using more accurate and consistent metadata. This paper presents the results of the initial phase of the project, in which a new metadata schema for the OLT's repository was developed through a systematic analysis of the collection, users' and prospective users' search needs, and the domain of HE learning and teaching. While the methods used to develop controlled vocabularies, such as subject thesauri, are well established, there has been far less discussion about how schemas for describing particular kinds of information resource should be constructed. This article contributes to this discussion by showing how methods used to build controlled vocabularies can be applied, and combined, to the development of a schema used to support effective access to a scholarly repository of national importance.