Advocates for the use of RDF as a model for metadata in the cultural heritage sector have frequently spoken of the death of the “record”. Indeed, the shift from a document-centric approach to one based on identified resources and atomic statements is an important one. Yet current work on validation as well as requirements for day-to-day metadata management and attribution point back to aspects of a record-driven worldview. This session will address some historical views of records, contrasting them with the formal model adopted by RDF 1.1 and commonly accepted best practices for Linked Data. Practical implications of the RDF model will be explored, with questions raised regarding the management of state, mutability, and “record” workflows. A provisional approach for managing RDF resources and graphs in record-like contexts is proposed, with connections to RDF Shapes, DC Application Profiles, and Linked Data Platform. Use cases from the Digital Public Library of America will be presented as illustrative examples.