In previous research, little social cognitional work has focused on the mental representation of the stereotype about compound social categories, which are defined as the intersection of two or more constituent categories and can be construed as subcategories. There are two models of social subcategory representation which can explain how to form the stereotype presentation of compound categories. Brewer's abstraction model and Smith' exemplar model can both be right, but maybe in different circumstances, because the stereotype presentation of compound categories may be influenced by familiarity, compatibility and some other factors such as perceiver' own category attributes. Thus, in present study, we explored the interaction among compound categories' familiarity, compatibility and participant' category attributes(in-group or out-group). We hypothesized that people would make exemplar-based trait judgments of social compound categories, consistent with Smith's model, when those compounds were relatively unfamiliar. The stereotype representation of compatible and familiar compound category might depend on abstract knowledge. Maybe there were differences in the stereotype representation between in-group and out-group memberships.We conducted three experiments in present study and improved priming paradigm to explore the issues above. In Experiment 1, we chose female kindergarten teachers as familiar compound category and male kindergarten teachers as unfamiliar compound category to test the effect of compound familiarity on the stereotype representation of the compound social categories. And in Experiment 2, we chose female secondary school teachers as the target group. In addition, "rational" and "exquisite" were respectively chosen as compatible stereotypical trait word and incompatible stereotypical trait word to explore the effect of compatibility on the representation of compound category' stereotype when the target was familiar. Based on experiment 2, male secondary school teachers and female secondary school teachers were recruited to explore whether there are differences in the response to compatible stereotypical trait word and incompatible stereotypical trait word between in-group and out-group memberships in Experiment 3.The results of Experiment 1 revealed that when the target was male kindergarten teachers, participants' response in description group was faster than that in definition group during recall tasks. However, there were no such differences in reaction time when the target was female kindergarten teachers. In other words, participants had exemplars of male elementary school teachers but had abstract presentation of female kindergarten teachers.The results of experiment 2 showed that participants activated the exemplars of male secondary school teachers when they rated incompatible stereotypical trait word "exquisite" in description group but when the stereotypical trait word was "rational", there was no exemplar activation. So the stereotype representation of male secondary school teachers relied on the stereotype compatibility of the subcategories(male secondary school teacher). The results of experiment 3 demonstrated that the effect of familiarity and compatibility on the stereotype presentation of compound categories is not moderated by in-group and out-group memberships.In conclusion, the stereotype representation of compatible and familiar compound categories depend on abstract knowledge, whereas when compound social categories are unfamiliar, perceivers can activate individual category members(i.e., exemplars). In addition, the stereotype representation of incompatible and familiar compound categories also depends on exemplars.