Realistic behaviors for computer-generated forces (CGF) are as crucial as realistic graphics and terrain to the creation of high-quality military simulations. While a variety of simulation-building and 3D-modeling tools exist to help with the construction of the latter, the development of CGF behaviors still typically requires programming code to be written, either in a standard language such as C++ or Java or in a custom scripting language. As a result, the subject matter experts (SMEs) with the tactical or operational knowledge about how CGF should behave are seldom able to directly specify that behavior for a simulation, because they lack the necessary programming skill. The researchers therefore set out to develop an approach to simulation behavior authoring that minimizes the amount of programming required while still allowing the creation of sophisticated behaviors. Two key observations guided this effort. First, there exist already a variety of largely visual "languages" for describing complex sequences of actions and conditions - such as flowcharts, finite-state machines, and decision trees - that are either familiar to or quickly understandable by non-programmers. Second, CGF behaviors, particularly at a tactical or operational level, can often be adequately specified using such lightweight procedural representations. The end result was a behavior authoring methodology that is founded on a lightweight, visual, procedural approach to modeling CGF behaviors. This methodology, which is embodied in a graphical editor and runtime engine, is intended to allow non-programmers to participate more directly in the behavior authoring process. It is also designed to encourage good development practices such as reuse and top-down design, to which end it borrows several elements of object-oriented programming, including the notion of behavioral polymorphism. This paper describes the basic authoring methodology and underlying behavior representation. Examples are drawn from the Counter-Strike simulation testbed constructed by the researchers.