Literacy testing has been researched as a social practice from different perspectives (McNamara & Roewer, 2006; Shohamy, 2001). Drawing on a Faucault inspired concept og governmentality in which literacy testing practices are seen as social technologies (Dean, 1999) and as a phenomenon closely related to supra- and transnational agencies this paper investigates the relation between state, pedagogy and conceptualizations of literacy. Drawing on data and findings from three ethnographic oriented studies of institutional testing practices of literacy in preschool, primary school and adult second language teaching in Denmark (Holm, 2004; 2007; 2009) this paper reveals the construction of values, ideologies and practices around institutional testing of litaracy in education. The analyses of testing instruments and assessment practices indicate among other things that testing of literacy have become a central instrument for the installation of the autonomous model of literacy (Street & Street, 1995) as the dominant conceptualization of literacy in post-modern national states, and that the state plays a more active role in this development than previously. Furthermore, the analyses reveal that the dominant testing formats are theoretically grounded in concepts that are insensitive to multilingualism and therefore often leads to negative categorizations of multilingual groups in society. Literacy testing is thus not only a question of measuring individual competences, but is closey related to broader social practices and values involving issues about identity, societal membership, citizenship and Bildung (Holm, 2007).