In this article, we provide a process description of scaffolding in music lessons based on the scaffolding model of Van Geert and Steenbeek (2005). Scaffolding is a form of socially mediated learning in which teacher and student constantly adapt their behavior to one another in order to reach a goal. To illustrate this process, data from video-observations of two teacher-student dyads in Suzuki string lessons were analyzed to see how well this data fits the Van Geert and Steenbeek model. Results show the dynamic connection between amount of instruction and instruction type on the one hand and performance level of the student on the other hand. Results also show that the process of socially mediated learning takes the form of an increasing overlap between student and teacher actions, i.e. an increase of the joint action over time.