Analyses of international policy cooperation are often marked by analytic and empirical confusion. First, by largely treating cooperation as a binary phenomenon (typically, as cooperation versus defection), they direct attention away from crucial issues of distribution, the possibility of suboptimal cooperation, and the degree of unrealized joint gains. Second, even when simple matrix games with known payoffs capture distributional conflict and Pareto-inferiority, they typically do so by suppressing the inherent uncertainty and the need to learn, especially with respect to payoffs and values. And, third, even when they take both power and knowledge-dependent joint gains into account, they often treat the two as competing alternatives or as analytically separable, rather than as inherently bound together in the bargaining process. This article describes the emerging negotiation-analytic approach and argues that it provides a useful framework within which these conceptual problems can be avoided and explanatory power can be enhanced. From a negotiation-analytic perspective, it argues, epistemic communities can be viewed as distinctive de facto natural coalitions of “believers” whose main interest lies not in meeting material objectives but, rather, in expanding to become winning coalitions capable of ensuring the adoption of specific policy projects. An epistemic community's actions can thus be understood as changing the perceived zone of possible agreement through well-understood ways that are favorable to its objectives. Although ultimately a community's influence is exercised through bargaining, there is practically no theory of bargaining elaborated in the epistemic communities approach. Despite this and other drawbacks, the approach helps account for players' interests and usefully insists on the importance of perceptions and learning in negotiation. The article concludes that the effects of shared beliefs and of policy conflict can be readily incorporated into the negotiation-analytic model of bargaining, thereby giving rise to more precisely drawn observations about the conditions under which “believers” will have the strongest impact on negotiated outcomes. This will in turn make it possible to improve policy prescriptions in the actual or potential presence of epistemic communities.