Managing the health of vehicle, crew, and habitat systems is a primary function of flight controllers today. We propose to develop an architecture for automating many of the response actions in problem situations and coordinating these actions with human response activities. Our approach is to develop a basic module that automates (1) Planning what response actions to take, (2) Executing these planned actions, and (3) Monitoring the effects of these actions to ensure they are successful. These modules will be organized into layers, operating at different levels of abstraction and with different time constants and time horizons. The architecture will support both hierarchical coordination between layers of control and peer-to-peer interaction among modules within a layer. This proposal is innovative in combining centralized planning for hierarchical coordination with distributed negotiation techniques for peer-to-peer coordination to provide flexible contingency response. The successful completion of a Phase II project will deliver a system health management architecture to NASA that improves system safety and makes more effective use of human time and capabilities when problems occur.