Distributed simulation is an approach to building large-scale simulation models from a set of independent simulator nodes communicating via a network. The U.S. Army uses distributed simulation systems for both training and analysis. Those systems include both crewed simulators and computer generated forces (CGF) systems; the latter use software, rather than human crews, to generate the behavior of entities in the simulated battlefield. CGF systems must include algorithms for all of the tactical behaviors that are needed for the simulation. One such tactical behavior is “Fire Zone Defense”. An algorithm for this behavior must select defensive deployment locations on the terrain for the individual entities (e.g., tanks) of a unit (e.g., a company) to effectively defend an assigned engagement area. The entities of the unit then move to those locations. We developed a new algorithm for the behavior. It combines a geometric terrain analysis algorithm, which creates a weighted graph representation of the terrain, with a greedy optimization algorithm that operates on that graph. The algorithm was compared experimentally with a previously existing algorithm for the same behavior. The comparison used a metric that measured the cumulative observation of the engagement area from the selected locations of the defending entities. Under that metric the new algorithm consistently outperformed the existing algorithm, with an average ratio of performance over 2. The execution speeds for the two algorithms were approximately the same.