The original JHOVE characterization framework is widely used by international digital library programs and preservation repositories. During the course of its extensive use over the past four years, however, a number of limitations imposed by idiosyncrasies of design and implementation have been identified. With funding from the Library of Congress under its National Digital Information Infrastructure and Preservation Program (NDIIPP), the California Digital Library, Portico, and Stanford University are collaborating on a two year project to develop and deploy a next-generation architecture providing enhanced performance, streamlined APIs, and significant new features. The JHOVE2 project generalizes the concept of format characterization to include identification, validation, feature extraction, and policy-based assessment. The target of this characterization is not a simple digital file, but a (potentially) complex digital object that may be instantiated in multiple files.