In this presentation, we explore the co-production of knowledge and power in secondary school physics classrooms. The basic premise is that the privileging of certain content in teaching has consequences for what the students are given the opportunity to learn, and can thus be regarded as an aspect of power (cf. Foucault 1982/2002; Öhman, 2010). The presentation will focus on the methodological considerations involved in analysing the co-construction of knowledge and power and outline the findings of our analysis. The empirical data consists of video recordings and field notes from physics lessons in three lower secondary schools (14-15 years old students), where the students are differently situated in terms of socioeconomic and cultural background. A key construct in our analysis is ‘governance’: we analyse power aspects in the teaching of physics by identifying actions that guide or direct other people's actions (cf. Foucault, 1982/2002). Thereafter, we investigate similarities and differences in the classrooms in terms of how governance is staged and what potential consequences this can have (see Danielsson, Berge and Lidar (2017) and Östman, Öhman, Lundqvist and Lidar (2015) for similar approaches used in science education). Teachers from all three schools adhere to a rather traditional interpretation of a physics curriculum, in that moral and political aspects are largely excluded. However, a more in-depth analysis highlights differences between the classrooms, in that the students in the three classrooms are given very different opportunities for participating in the teaching and learning, and creating relationships with the content. For example, in two of the studied classrooms, the teacher to a large extent controls the content progression, but in one classroom this takes place by inviting students to contribute physics knowledge that has not yet been presented, whereas in the other classroom it takes place by asking questions of a controlling character (thus, checking that they have followed what just have been said). Hence, the conditions for taking part in knowledge-making in the classrooms vary greatly. In the context of this symposium, we are interested in discussing how the production of categories of difference (such as social class and gender) can be taken into account in an analysis of didactical interactions, in ways that highlight potential inequalities without reproducing those through the analysis.