The simulation and training industry has decades of experience in specifying and acquiring aircraft flight simulators for aircrew training. This experience, however, does not translate well to deriving requirements for the relatively recent acceptance of virtual maintenance trainers into the industry. In designing a Virtual Maintenance Trainer (VMT), requirements need to be based on the desired level of training for the practical application of the knowledge and skill obtained from the VMT. The aircraft maintenance training continuum encompasses the knowledge domain from the ab initio aircraft maintenance technician to the 20-year seasoned master mechanic. The learning objectives across this continuum are very different; a "one size fits all" set of maintenance training requirements result in unnecessary costs and/or ineffective training scenarios. Today's VMT technology can support a wide variety of requirements regarding procedural fidelity, graphical fidelity, and interactivity methods. Procedural fidelity establishes if a single trainee interaction event is required to remove and replace an entire component, or if the trainee needs to select every individual nut/bolt/washer to complete the lesson. Graphical fidelity ranges from a hand modeled artist renditions of the devices to full CAD data translation of the aircraft systems. Interactivity methods cover the spectrum from a mouse click to fully immersive 3D haptic gloves in a stereoscopic system. The scope of requirements for a specific VMT application should be driven by the specific needs in the training continuum. This will maximize training value while minimizing training costs. This paper reviews the requirements from three different large-scale VMT programs: the Canadian CH-147, Australian F/A-18E, and F-35 JSF. It highlights how the differing requirements are derived to meet the desired training level within the maintenance training continuum. Through this we seek to describe the impacts requirements levy on the cost and complexity of the resulting VMT.