Understanding the relationship between level of fidelity and training effectiveness is a question repeatedly posed in the training literature in previous decades. Specifically, to what extent does increasing the fidelity of a training system improve or not improve training effectiveness. As research aimed at answering this question is aggregated in order to conduct larger meta-analyses, a problem arises that affects both researchers and acquisition professionals alike. That is, the training community lacks an accepted, standardized approach to decomposing and defining fidelity in such a way that is generalizable across training effectiveness evaluations and requirements documents. For the researcher, this gap creates an issue when attempting to aggregate the results of multiple training effectiveness evaluations when each of the various authors is using their own operationalized definition of high and low fidelity. For the acquisition professional, this gap creates an issue when attempting to develop fidelity requirements for future training systems or when attempting to define the fidelity of existing training systems. This paper outlines the development of a novel approach to solving this issue using the Simulation Fidelity (SiFi) scale. The SiFi scale is a 6-point scale used to rate the fidelity of a simulation. Ratings from the SiFi scale can be used to construct a descriptive fidelity model (i.e. the level of fidelity of an existing system) or to construct a prescriptive fidelity model (i.e. the required level of fidelity for a future system). In addition to providing real-world examples of the applications of the SiFi scale, previous empirical evaluations of the scale are discussed. The SiFi scale and approach offers a standardized means of defining training system fidelity in the context of training effectiveness research as well as defining and developing requirements for training system acquisition.