Most studies of civil society are mainly normative, both in their conceptualization and in their theoretical assumptions. We suggest a new, or better revised, research agenda of the study of civil society, which should include (at least) these four points. First, the concept of civil society should be seen as a heuristic device. Second, the separation between ‘civil society’ and ‘uncivil society’ makes sense only in a normative framework, and then exclusively with uncivil society defined as a sub-set of civil society. Third, empirical research on civil society should study the nature of the relationship between civil society organizations and democracy/democratization, rather than assume it. Fourth,‘uncivil’ movements and contentious politics should be included in the study of civil society.