The normative aim of childhood studies is to show that children are and should be recognized as active shapers of their lifeworlds. In this article, we discuss which concept can best be used to accomplish this. Our thesis is that the agency concept ubiquitous in childhood studies only inadequately advances the field's normative agenda. Mostly containing some hidden normativity, its meaning remains primarily descriptive. Indeed, children always have some kind of agency, regardless of the conditions they live in. They may exercise agency while still being manipulated or otherwise rendered incapable of acting as autonomous human beings. Against this backdrop, we first delineate the deficiencies of the notion of child agency and try to show why it should be replaced by that of autonomy in order to preserve and make explicit its hidden normative impetus. Second, we seek to clarify which understanding of autonomy is able to fulfill our aspirations. We oppose individualistic notions of autonomy and, by the same token, draw on criticism of the insufficient attention often paid to structural social factors. Eventually, we aim to develop a social concept of child autonomy that takes into account the relationality of human existence and replaces the asymmetrical relationship between children and adults with an equal appreciation of differences.