Role-plays are a common pedagogical tool in the Social Sciences. As an imitation of societal practices, role-plays are thought to support the development of argumentation and decision-making skills among learners. However, argumentation and decision making are also goals in science education in general and in socioscientific issues-oriented science teaching in particular. This paper discusses a grounded theory (GT) approach to evaluating students' performance within role-playing exercises. The context is climate change. Data come from 4 different role-playing scenarios covering climate change which were developed in parallel for Biology, Chemistry, Physics, and Politics education. Role-plays in 20 different learning groups were videotaped (5 per subject). An evaluation pattern was developed step by step according to GT. Finally, graphic representations of all the role-plays were derived. The representations enable a quick overview of the role-plays and allow the identification of four basic types of role-playing: role-plays that are (1) completely directed by a group of student moderators, or (2) by the teacher, (3) medium-quality role-plays with a certain amount of interactivity and free argumentation, and (4) real, spirited debates. Implications for the use of role-playing exercises in science education are derived, including the induction of such role-plays through the use of role cards and the influence exhibited by teacher behavior.