For decades, computer simulation professionals have taken numerous routes to improve the performance of their simulation systems.These measures have included, but are not limited to improved data and behavior representations, advanced development architectures, and enhanced graphics technologies. Another remarkable concept used to improve the performance and capabilities has been the use of distributed computing techniques, dubbed distributed simulation. A new and potentially high impacting expansion of distributing simulation involves tying that methodology with that of selective fidelity, a concept that allows a simulation system to modify its own level of fidelity based on what aspects of the simulation are important or interesting to a user or observer. Using multiple simulation systems tied together by a run-time infrastructure (RTI), the software implementation of the High Level Architecture (HLA), a selective fidelity federation can be implemented by changing ownership of simulation objects between federates, depending on what aspects of the simulation are important. This paper describes the theoretical and implementation details involved with building such a system. Included in this description are explanations of the concepts of fidelity and selective fidelity, a summary of HLA and the RTI, a detailed account of the architecture necessary to implement such a scheme, and a section predicting possible applications for such an idea.