Part of the solution to any problem is the correct specification of that problem. Debate surrounding the agent–structure problem within IR theory has become confused because it is not always clear that the participants in discussion of the issue are talking about the same problem. Questions that are considered to be epistemological by one contributor are believed to be methodological by another. Ontological issues are regularly confused with matters of explanation and there is widespread confusion about just what the problem is. The aim of this chapter is to identify what lies at the heart of the agent–structure problem and disentangle this from the other issues that surfaced during the debate surrounding this issue within IR, but which are not an integral part of it. It is important to examine the nature of the agent–structure problem in order to ascertain whether the contributors to the debate are addressing the same problem. This clarification gains added significance if we are to remain clear about the distinctions, if any, between differing forms of problem – levels-of-analysis problem, agent–structure problem and macro–micro problem – and about what they entail. It is also necessary because it is the only way of evaluating the arguments for and against particular points of view: we can only understand why some theorists have advocated certain approaches and others objected to them, and judge whether their arguments are sound, if we know what the problem is and what it is they are advocating/objecting to.