Political scientists are paying increasing attention to the effect that shared membership in intergovernmental organizations (IGOs) has in international politics. A number of studies have examined the role that shared membership in IGOs has on dependent variables such as conflict, trade, interest convergence and the diffusion of human rights norms. More recently, scholars have turned their attention to explaining the variation that exists in the extent to which states join IGOs in the first place. In this paper we advance this literature by adopting a network theoretic perspective of IGO membership. Rather than considering the IGO network as simply a collection of ties between states, we consider the ways in which the IGO network can be conceptualized as a number of distinct communities that consist of states and IGOs. We posit that accounting for membership in these communities allows IR scholars adopt a more nuanced understanding of the causes and effects of IGO membership. Our argument is that, depending on the logic of IGO joining, we would expect these "clubs of clubs" or IGO communities to be defined on differing grounds. In the empirical part of the paper we use the network analytic tool of modularity maximization to detect the IGO communities in the global network for the period 1950-2000. We describe how the IGO communities have developed over time and test the extent to which factors such as development, geography, regime type, alliance ties, language, religion and colonial ties explain the IGO community structure.