Department of Defense (DoD) systems are often highly complex, costly and have extraordinarily long life cycles. Due to these characteristics requirements that these systems will need to meet over their life cycle are highly uncertain. To meet future requirements more rapidly at a lower cost requires an understanding of how to manage uncertainty and architecture to make these complex systems more flexible, adaptable and affordable. This paper proposes an alternative approaches to traditional development through managing uncertainty and architecture in an iterative fashion with decision analysis methods. Several specific methods and tools are discussed to include: Influence Diagrams, Design of Experiments, Design Structure Matrix and Target-Oriented Utility. Collectively the approach identifies the component and architectural drivers of cost in military systems.