This paper describes techniques to augment the behavioral models of automated Opposing Forces (OPFOR) Individual Combatants (ICs) with a realistic, practical set of weapon firing behaviors for virtual Military Operations in Urban Terrain (MOUT) training environments. These behaviors represent a formalization of target acquisition and firing execution tactics and techniques based on doctrine, input from subject matter experts, and observations from the field. The formalisms are based on Behavior Transition Networks (BTNs), an extension of Finite State Machines (FSMs). A COTS toolkit allows for rapid visual behavior specification, testing and modification, easy simulation integration, and flexible architectures. The behaviors are designed hierarchically so that the actions and goals of a human combatant can be modeled at various levels of detail. Polymorphism is used heavily to alter the behaviors based on the type and current state of the IC at all levels of the model. For example, an IC just exposed to a stun grenade behaves very differently from one who has not been so exposed. A motivated, well-trained, elite OPFOR IC behaves very differently from a conscript. Uncertainty is incorporated both in the initial selection of attributes (boldness, speed, aiming accuracy, etc.) of the ICs to give their behaviors natural variation and in runtime execution of decision making to keep even a single IC from being too predictable. This paper also describes the behavior modeling process itself from knowledge engineering to formalization and implementation through validation. The initial prototype is described in which IC behaviors are implemented and interfaced to a real-time IC simulation.