The complexity of the policy process is such that analysts often resort to metaphorical representations of its most salient aspects. Sometimes these metaphors are used deliberately but, in most cases, they are implicitly built into their theoretical frameworks. This article argues that commonly used metaphors based on the paradigmatic notion of 'control' have ceased to be relevant to the analysis of contemporary policy dilemmas. Two new conceptions of the policy process have emerged from the new sciences of complexity. Both chaos theory and models based on the concept of ,organizational closure' clearly reveal the self-organizing logic inherent in the problems confronting managers and policy-makers today. The main focus here is on examining the rationales for, and the potentials of, metaphors derived from these paradigmatic innovations -innovations which can be situated within an emerging postmodern culture insofar as they emphasize indeterminacy and the role played by social actors in constructing the social situations in which they find themselves. It is also argued, however, that within very specific contexts the notion of control may still be valid.Neither policy analysts, nor social theorists, nor ordinary mortals can avoid using metaphors -sometimes called 'models' -when talking about public affairs. Metaphors are more than superficial analogies used merely for stylistic purposes; they serve to transfer certain relationships from one level to another where such relationships rev.eal logical connections that had remained unnoticed (Black, 1979). This intrusion of rhetoric in the policy sciences is sometimes viewed as an unfortunate obstacle to the articulation of a truly 'scientific' language. But, as it is becoming more widely recognized, rhetoric and policy analysis are inseparable (Majone, 1989;Throgmorton, 1991).This being granted, two problems arise: First, how do we know the metaphors we are using are adequate. Second, how do we guard against the risk of confusing images of an elusive reality with that reality itself; in other words, how can we avoid what A. N. Whitehead called 'the fallacy of misplaced concreteness?' This paper examines the development and relevance of metaphorical construction in systems-theoretic approaches to policy analysis and draws several inferences from this particular case for the use of metaphors in the policy sciences as a whole.During the last two to three decades, systems theory and cybernetics have proven to be a fertile source of ideas for the many disciplines that are constitutive of the policy sciences, e.g., organization theory and information management, institutional design, strategic planning, budgeting and policy * The author wishes to thank Michael Howlett for his helpful comments on an earlier draft.