Over the past 20 years, the participation of laypersons or representatives of civil society has become a guiding principle in processes of research and innovation. There is now a significant literature discussing collaboration between civil society organisations (CSOs) and researchers, with two interesting gaps. Firstly, the fact that research is mainly conducted within projects is often underestimated, although the format significantly frames knowledge production. Secondly, researchers and civil society organisations are closely related to their respective communities. We argue that this constellation – of project-related format, in combination with a strong relationship to communities – results in conflicts that express and lead to identity politics. The analysis is based on conceptual considerations as well as empirical findings, which were developed within the EC-funded CONSIDER project (2012–2015). It can be shown that identity politics is performed by socio-epistemic tactics, which are used to order the socially as well as epistemically hybrid space within projects. To explain differences in conflict intensity, we suggest the distinction between weakly tied and strongly tied identity politics. In sum, identity politics can be seen as one key element for social as well as epistemic control in transdisciplinary research projects.