The delegation of power and tasks to unelected bodies, a pervasive trend in modern democracies, raises a democratic puzzle. These institutions are removed from majoritarian democratic procedures, breaking the "chain of command" from citizens to elected officials and administrations. Still, they make claims to political authority and legitimacy. This dissertation asks: Under which conditions is the delegation of power to unelected expert bodies democratically legitimate?The dissertation focuses on European Union agencies. It has been common to justify the power of EU agencies by appealing to what the dissertation calls technical legitimacy. The argument says that political neutrality and technical expertise are independent sources of legitimacy for unelected bodies. Certain institutions are legitimate despite, or even because of, their isolation from majoritarian democratic procedures.Through an introductory chapter and four articles, this dissertation aims to delineate the boundaries of technical legitimacy. It cautions against too much faith in technical legitimacy, for instance through the widespread claim that agencies' value-freedom will ensure their legitimacy. But it also disagrees with the position that technical legitimacy is never appropriate. The challenge is to figure out when technical legitimacy may be the appropriate standard-and what should come in its place where it is not.The dissertation combines empirical and normative analysis. A mixedmethods media analysis finds that the legitimation arguments used about agencies in public debate depend on their scientific "hardness" and public salience. A case study of the European Border and Coast Guard Agency, Frontex, uncovers how appeals to technical legitimacy were central when it iii iv summary received a new mandate in 2016. On the normative side, the dissertation argues that agencies face different legitimacy demands depending on their level of epistemic uncertainty and the consequences of potential errors. It also develops a symmetry criterion for legitimacy. The legitimacy of an agency might depend on the level of power delegated to other agencies in a policy area, so that the system as a whole leaves no gaps. Taken together, the dissertation argues against a one-size-fits-all approach to legitimacy: Every institution must not satisfy everything that democracy demands, but the system as a whole must. I want to thank my supervisors, Erik O. Eriksen and Asimina Michailidou, for all discussions and comments on my drafts and ideas. It's been a privilege to work with Erik Oddvar. His commitment to the combination of normative and empirical analysis has shaped my project in more ways than one, and his advice always gives me something to think about. Asimina's perceptive readings and constructive feedback have been frequent lifesavers. Co-writing an article with Asimina was a particular joy and motivation. arena Centre for European Studies at the University of Oslo has been my academic home since I started as a part-time research assistant more than six y...