An elusive goal in virtual training environments is to be able to dial up the opponent of choice – e.g., the Iraqi Republican Guard, an Hamas-type of Suicide Bomber, or the clandestine minions of Bin Laden, as a few examples. In researching alternative ways to offer such a "dial up" capability, our focus thus far is to analyze actual organizations to identify "individual differences" in the form of Performance Moderator Function scorecards and a hierarchical game theoretic approach that captures the situation, organization, population, ideologic/motivation, strategic, and tactical layers of their decision making. We are also crafting a tool that can use the scorecards to semi-automatically assemble and deploy non-traditional Semi-Automated Forces or agents on a virtual battlefield. As an initial proof of concept test, we have manually applied the approach to a scenario involving a bank bomber approaching a vehicle checkpoint. The results to date indicate the approach seems to be a useful representational formalism for generic, implementation-free models of terrorist organizations and the behavior of their members. Our next steps will be to scale up the approach and try to implement it as a terrorist generator for an existing virtual-reality training environment.