"Fragile states" is one of the latest buzz-phrases to emerge within the aid sphere, and thus of increasing importance to those working with international educational development today. Buzz-phrases and buzzwords play a central role in international aid policy because they help frame solutions by suggesting that problems can be clearly defined, categorised, and tackled in a systematic manner. I present findings from a qualitative study examining how the terms "fragile states" and "fragility" are defined, conceptualised and used by education aid professionals at the global policy level. The dissertation also explores the broader social context—the international aid community as a whole—in which these terms emerged. My conceptual framework builds on critical discourse analysis as conceived by Norman Fairclough, which draws on systemic functional linguistics, and envisages a three-dimensional model of discourse. Using semi-structured interviews and an in-depth review of agency documents, I analyse the discourse of a small but influential group involved in shaping global policy related to education and fragility: the Inter-Agency Network for Education in Emergencies Working Group on Education and Fragility. I demonstrate the lack of a consensus on what the terms "fragility" and "fragile states" encompass and exclude, and argue that these terms are confounding and stigmatising, and thus of limited utility, beyond bringing a few more countries onto the international aid agenda and stakeholders around the same table. The dissertation posits that these terms as they are currently used are more harmful than helpful, and are potentially steering the international aid agenda in an unfortunate direction and makes a case for the importance of well thought-out linguistic interventions in improving aid policy in practice. By investigating the emergence and current usage of these terms, I aim to encourage self-reflexivity among aid practitioners, and to contribute to a clear, shared, and useful understanding of the concepts these terms are meant to denote. Such clarity, I argue, will not only increase the opportunities for effective programming, but could also prevent "fragility" from becoming merely another superficial and problematic stereotype used to identify the so-called "developing" or "third world" countries.