Abstract The concept of ‘scaffolding’ introduced by Wood et al. (1976) figures prominently in educational research but lacks the empirical rigour that allows researchers to establish whether or not teacher assistance to students is an instance of scaffolding. We used conversation analysis to provide an empirical basis to the notion of ‘responsiveness’ (contingency) that Wood et al. treat as a fundamental characteristic of scaffolding. We analyzed dyadic teacher–student interactions in Dutch 1st grade secondary school mathematics classes and developed responsiveness as an interactional phenomenon: the concept has to rest on the analysis of how the learner's actions and the tutor's responses are interactionally brought about.