In May 2005, the PREMIS Working Group released the first version of the Data Dictionary for Preservation Metadata: Final Report, a community consensus based preservation metadata standard. PREMIS XML schemas were also published to support the implementation. Since then, many organizations started implementing PREMIS in their repositories, during which a handful of common implementation issues surfaced. The community is finding a variety of solutions to these issues and anticipates the emergence of best practices. The PREMIS Implementers' Group (PIG) was formed after the release of the data dictionary to address implementation issues through open discussions and consensus building processes. In August 2006, the PREMIS Editorial Committee was established to coordinate and approve future revisions of the Data Dictionary and XML schema, as well as provide guidance for implementation. This paper summarizes some of the experiences gained in implementing PREMIS in container formats, especially in the context of the METS and MPEG-21 DID frameworks. Attempts are made to draw best practices from our experience for future implementers.