EITHER the traditional realist explanation nor liberal and constructivist alternatives are adequate to explain NATO’s formation. Existing explanations of the formation of NATO in International Relations (IR) theory all begin from the position that explaining NATO is a matter of explaining the specific decisions made by individual state actors. It is self-evident to most IR scholars that interstate alliances are arrangements produced by what states do, and that “action” in world politics means “state action.” This reductionist explanatory position parallels a number of individualist stances in the wider sphere of social theory, drawing its intuitive plausibility from a posited equivalence between individuals in society and states in international society: states are, as it were, the “people” of international society. 1 But by adopting this reductionist stance, IR theorists also acquire the central explanatory weaknesses of an individualist approach, particularly inasmuch as they aim to combine an individualist focus with a desire for explanations of social outcomes based on necessity rather than contingency. This “necessity individualism” hampers existing accounts in their efforts to explain the existence of the Alliance. In particular, two explanatory weaknesses characterize necessity individualist accounts. First, such accounts are characterized by a tendency towards the teleological reconstruction of history, in that they tend to read the stable bipolar situation of later years backwards into the incredibly ambiguous period of the period immediately following the Second World War. Second, necessity individualist accounts do not contain an adequate notion of agency, and as such deny the creative aspects of social action. As is usual in social theory, the empirical weaknesses of existing accounts are generated by their underlying theoretical problems; therefore, these theoretical problems must be addressed in order to generate a more robust empirical account. As necessity individualism is the underlying problem, it must be replaced with different theoretical