Emotion education is enjoying new-found popularity. This paper explores the 'cosy consensus' that seems to have developed in education circles, according to which approaches to emotion education are immune from metaethical considerations such as contrasting rationalist and sentimentalist views about the moral ontology of emotions. I spell out five common assumptions of recent approaches to emotion education and explore their potential compatibility with four paradigmatic moral ontologies. I argue that three of these ontologies fail to harmonise with the common assumptions. Either those three must therefore be rejected or, if we want to retain one or more of them (for instance, Jesse Prinz's recent rebranding of hard sentimentalism that I explore in detail), we need to revise our assumptions about the practice of emotion education in ways that are both radical and, I argue, ultimately unacceptable.The tenuous relationship between theory and practice continues to haunt practical disciplines. A psychotherapist uses cognitive behavioural therapy while questioning behaviourism as a psychological model. A school administrator abhors the subjectivist definition of quality underlying Total Quality Management but still decides to apply its practical tools and techniques. An educational practitioner sets out to develop a 'praxis' of teaching irrespective of any established theory. Are these better understood as cases of undue inconsistencies or of healthy pragmatism? There is much talk among academics of 'guilt by association', where reservations about underlying theories carry over into related practices or vice versa; there is less mention, for some reason, of 'pride by association', although that seems to be the other side of the same coin.