In this study, we tested the theoretical validity of both Social Identity Theory (SIT) and the Bounded Generalized Reciprocity Hypothesis (BGR) for explaining in-group cooperation in real social groups. While most previous investigation on real social groups has found support for SIT, confirmatory evidence for BGR remains limited. We conducted a vignette experiment in which reciprocity was manipulated by controlling knowledge of group membership. The participants were 117 undergraduate students who were baseball fans. To control expectation of reciprocity in the experiment, we asked participants to imagine scenarios where they helped others who wore either the same baseball team T-shirts as they did or plain T-shirts. Consistent with BGR, participants tend to cooperate with in-group members when both they and their partner knew that they might be supporters of the same team. On the other hand, participants showed in-group cooperation even when only they knew their partner’s group membership. This finding coincided with SIT. These results thus further verified the theoretical validity of both SIT and BGR in real social groups.