The use of controlled vocabularies in bibliographic metadata is essential to ensure interoperability in terms of data exchange and the provision of value added services. Repositories are using controlled vocabularies (e.g. from the Dublin Core Metadata Initiative) and application profiles (e.g. info: eu-repo) across system boundaries. However, in practice, one can observe a lack of consistent use, granularity and regular maintenance which has significant negative impact on resuability of metadata. In addition, new application scenarios arise (e.g. recording of funder and project/grant information, linking publications with other scholarly output, producing reliable usage statistics and sharing of scholarly information across different systems, like literature repositories, data repositories and current research information systems) and thus increased demands on controlled vocabularies. The COAR Interest Group on “Controlled Vocabularies” aims to address this problem and is developing a set of controlled vocabularies for open access repositories as a community-driven effort. Using the example of the COAR ResourceType vocabulary to describe the genre of a digital object organizational, methodological and technical steps of its implementation are presented.