Several levels of systems analysis are required to implement the model in sufficient detail to support decisions on various countermeasures. First, an overarching model must bring together the mass of information on different types of threat scenarios, different groups of perpetrators, their objectives, and the damage they can cause. Second, there must be an analysis of the various potential targets, including infrastructure systems and networks. Third, there must be an assessment of the consequences of the various attack scenarios. This paper focuses on the overarching model (the first level), with the aim of supporting decisions and setting priorities among homeland defense countermeasures. The pilot model presented is not a finished product, but rather a "blue print" for a global, more detailed model that could be used and updated in real time to support protection decisions. The overarching model is based on the engineering risk analysis method. The first step in the analysis is to combine probabilistic modeling of the actions of various terrorist groups with an assessment of their objectives and the consequences for American interests. The key variables of the overarching model are the various groups or individuals who can be potential perpetrators; the objectives of these groups; the means at their disposal; the nature of the potential threats; the different classes of targets; and the means of delivery. One key assumption of the model is the use of the rational decision analysis model in a descriptive mode. The data used in the model represent the beliefs of American experts regarding the probabilities of actions and the value systems of the various groups of perpetrators. In the dynamic/game-theoretic stage described in this paper, each side in the game puts a probability distribution on the beliefs of the other. The authors illustrate the general way in which the probability and consequences of an attack are computed and then show how to compute the benefits of various countermeasures in a more elaborate model in which the realizations of each random variable and event are more complex. Extensive figures, 14 references, and appended supplementary information and data