Continuous professional development of teachers, fostered by collaborative knowledge sharing, is a pivotal element in providing high quality education. In this context, Networks of Practice (NoP) have been proposed as a beneficial digital environment where such a process can be enhanced. Additionally, collaborative tagging systems (CTS) have gained increasing popularity, allowing individuals to collaboratively contribute to a growing repository of (online) resources that can improve teaching practice. However, CTS systems are seldom dedicated to teachers. Even more so, only limited evidence is available on whether and to what extent teachers actually engage in (online) knowledge sharing activities, as they are largely supposed to be preparing their classes in isolation (Hou, Sung, & Chang, 2009) . The present study addresses this issue by providing empirical evidence on a CTS, which aims at establishing a NoP among (German speaking) primary and secondary school teachers. We collected longitudinal data on individuals’ sharing behavior on the CTS in question. Second, we employed social network analysis (SNA). Building upon the work of scholars like Halpin and colleagues (2007), we considered a “tripartite graph structure within tagging systems” (p. 213). Focussing on individuals’ tagging behaviors, we simplified the network into a one-mode projection concerning only users, and a two-mode projection concerning users and tags. In order to analyze the type of resources that have been shared, we ran a Clauset-Newman-Moore cluster analysis to identify subgroups among the users that link via commonly used tags. The underlying data was collected from the 1 st of September 2013 until the 31 st of August 2014. The statistical software package R was used to collect and analyze the data. Our results indicate that 244 users added at least one resource with a minimum of one tag. This resulted in 3745 shared resources and 3218 unique tags being used. A subset of isolated users (41) used tags that were not used by any of their colleagues. In contrast, more than 50 percent of teachers used at least 16 tags that were commonly used among participants of the CTS. Our cluster analysis revealed 4 clusters, whereby the isolated users were put into one cluster. The shared resources covered the general topic domains of “social media“, “open educational resources” and “(multi)media learning resources”. These findings provide valuable insights. First, our results suggest that teachers actively engage in the CTS to store and exchange information and resources, thereby contributing to the formation of a NoP among teachers. Second, using SNA allowed us identifying network structures and reveals (thematic) commonalities among users that might have otherwise been overlooked. Overall, future research can build upon these findings by investigating the possibility of creating spill-over activities (e.g. face-to-face workshops, tailored information channels) among sub-groups of alike-minded users.