The term “digital repository” is used in disparate contexts, in both in a formal sense to describe a complex and complete preservation system and its policies, and in a more intuitive but less clearly defined sense, often implying simply a robust storage system, or a content management system. Lack of formalism in use of this term can impede development of formal requirements for building specific repository systems and, consequently, the widespread use of such systems. On the other hand, the complexity of a formal, system-wide view can impede understanding and implementation in many environments.To attempt to reconcile these concerns, we present a pragmatic definition and scope for digital repositories (systems whose primary function is the long-term preservation of digital objects). Taking a “bottom-up” approach that builds abstractions on top of reliable storage systems, we establish a minimal general vocabulary for the expression of digital objects, independent of policy, in the form of a low-level content model. We show that the “primitives” supplied by this vocabulary can be used to express the particular content and metadata models for several disparate repository case studies.