This paper presents an automated model for generating courses of action in support of an Operations Other Than War (OOTW) simulation. The model simulates the decision making of a theater level staff in the OOTW humanitarian assistance mission environment. The model uses probabilistic forecasting models and Bayesian techniques to predict what the state of a region in the theater will be some time in the future. Decision tree structures and the forecasting module are used to solve the decision making problem using expected utility. The model uses pairwise comparisons of utility attributes to obtain a decision maker's preference structure, This structure is applied over a multi-attribute utility function and the decision tree. to find the optimal course of action for some region of the theater at a specific time. Some variations on Lanchester's attrition equations are used to model attrition, the effect of civilians in a combat zone, and the effect of rules of engagement. The model was tested using data representative of Somalia in late 1992. The results indicated the best approach in this instance is to initially provide a high level of aid to reduce the civilian starvation rates then transition to a more aggressive posture with a strong force in readiness to retaliate for attacks by opposing forces.