In modern warfare the changing tactics of asymmetric threats present an ongoing need to disseminate lessons learned straight from the battlefield to a wide audience of personnel. Interactive virtual environments have been shown to be effective for training, and distributed game-based architectures contribute an added benefit of wide accessibility. However, conveying new knowledge with interactive training requires either the development of new simulation behaviors or the availability of training personnel for role-playing functions. This presents a constraint on either the speed or breadth of concept dissemination, but one which can be circumvented with virtual training demonstrations. Demonstrations have been favored by the Army as a complement to more traditional training materials because they accelerate learning, stimulate interest, and communicate better than text. They also can be delivered on a wide variety of hardware platforms and accomplish almost instantaneous shared knowledge. Unfortunately, demonstrations have received little attention in the research literature and there is little consensus on what constitutes a good demonstration. We describe two parallel avenues of research towards the rapid construction of effective demonstrations. The first avenue's goals are to: clearly articulate the nature and purpose of demonstration; compare related areas of research (e.g., observational learning studies, behavioral modeling training) to identify factors influencing demonstration effectiveness; and define a set of component capabilities, guidelines and best practices for creating effective demonstrations. The results inform the second avenue's investigation of how a demonstration authoring toolset can be constructed from existing virtual training environments using 3-D multiplayer gaming technologies. We then outline four potential uses for our work, specifically geared toward authoring demonstrations for Army curricula.