Health insurers increasingly compete based on their networks of covered medical providers. Using data from Massachusetts’ pioneer insurance exchange, I show evidence of substantial adverse selection against plans covering the most expensive and prestigious academic hospitals. Individuals loyal to the prestigious hospitals both select plans covering them and are more likely to use these hospitals’ high-price care. Standard risk adjustment does not capture their higher costs driven by preferences for using high-price providers. To study the welfare implications of network-based selection, I estimate a structural model of hospital and insurance markets and use the model to simulate insurer competition on premiums and hospital coverage in an insurance exchange. I find that with fixed hospital prices, adverse selection leads all plans to exclude the prestigious hospitals. Modified risk adjustment or subsidies can preserve coverage, benefitting those who value the hospitals most but raising costs enough to offset these gains. I conclude that adverse selection will encourage plans to limit networks and star academic hospitals to lower prices, with the welfare implications depending on whether those high prices fund socially valuable services. * Email: email@example.com. Website: http://scholar.harvard.edu/mshepard. Click here for the latest version. I thank my advisors David Cutler, Jeffrey Liebman, and Ariel Pakes for extensive comments and support in writing this paper. I thank the Massachusetts Health Connector (and particularly Michael Norton, Sam Osoro, Nicole Waickman, and Marissa Woltmann) for assistance in providing and interpreting the data. I also thank Katherine Baicker, Amitabh Chandra, Jeffrey Clemens, Keith Ericson, Jon Gruber, Kate Ho, Sonia Jaffe, Tim Layton, Robin Lee, Greg Lewis, Tom McGuire, Joe Newhouse, Daria Pelech, Amanda Starc, Karen Stockley, Rich Sweeney, Jacob Wallace, Tom Wollmann, Ali Yurukoglu, and participants in the Harvard Industrial Organization, Health Policy, and Labor lunches for helpful discussions and comments. I gratefully acknowledge data funding from Harvard’s Lab for Economic Applications and Policy, and Ph.D. funding support from National Institute on Aging Grant No. T32AG000186 (via the National Bureau of Economic Research), the Rumsfeld Foundation, and the National Science Foundation Graduate Research Fellowship.