© Springer-Verlag Berlin Heidelberg 1995.This paper describes the design and implementation of a layered agent architecture for decision support applications in general and for distributed medical care in particular. Three important characteristics which shaped the agent design are identified: distribution of data and control, information uncertainty, and environment dynamism. To provide appropriate decision support in these circumstances the architecture combines a number of AI and agent techniques: a symbolic decision procedure for decision making with incomplete and contradictory information, a concept of accountability for task allocation, commitments and conventions for managing coherent cooperation, and a set of communication primitives for inter-agent interaction.