This article explores the constructivists' institutional socialization hypothesis, positing that intergovernmental organizations (IGOs) make member-state interests more similar over time, thus promoting interest convergence. We first show how this hypothesis can be tested systematically using relatively new data on dyadic interest similarity and joint structured IGO membership, and then we conduct a series of empirical tests. Our results show strong statistical support for the institutional socialization hypothesis, using both global and more restricted regional samples. We also demonstrate how our results are consistent with a longer-term socialization process and cannot be explained by the short-term effect of institutional information. Finally, we show some limits to the institutional socialization hypothesis. Unstructured IGOs reveal no effect in promoting member-state interest convergence. Following recent theory arguing that great powers in the international system often use IGOs for coercive means, we find that institutional socialization gets weaker as the power imbalance within the dyad grows.Thanks to Chuck Boehmer, Heather Elko McKibben, Kate Floros, Erik Gartzke, Chuck Gochman, Michael Goodhart, Yoram Haftel, Volker Krause, Dan London, Andrew Long, Ed Mansfield, Lisa Martin, Tim Nordstrom, Zeki Sarigil, Meg Shannon, Dan Thomas, Lora Viola, Basak Yavcan, and two anonymous reviewers for data, comments, and/or helpful suggestions.