Current Institutional Repository packages do a poor job of maintaining the article’s metadata in a consistent fashion. Documents and other entities are unreliably identified and there exists no mechanism for correlating related data between multiple repositories. A consistent reference service (CRS) mediates and maps between different identifiers, from multiple sources. It overcomes the shortcomings of packages such as EPrints and allows the construction of useful applications and services, such as automatic CV generation or citation impact profiling. This project has developed a highly efficient and scalable CRS, capable of tracking many thousand identifiers. It utilises semantic web technologies to remain open and responsive, providing intuitive and flexible services for searching and retrieving information. A sophisticated plug-in for the EPrints software has been developed, which utilises the CRS to improve the inherent consistency of the metadata; reinforce the use of local naming schemes and significantly enhance the repository’s user interface. A CRS deployment is already in active use by researchers of the ReSIST Project.