An autonomic system is structured as a network of autonomic elements that collaborate to achieve the system?s purpose. This paper examines the potential benefit of using well-established systems concepts and techniques in the development of such systems. In particular, it considers the possible role of Checkland?s Soft Systems Methodology and Beer?s Viable Systems Model in system design. The paper summarizes the relevant aspects of each approach and then assesses both their individual and joint strengths in support of the construction and evaluation of designs. Some practical issues in the use of these approaches are also identified. The discussion is illustrated using aspects of the design of an autonomic operating system.