Two studies tested the conditions under which social environments can undermine automatic gender stereotypic beliefs expressed by women. Study 1, a laboratory experiment, manipulated exposure to biographical information about famous female leaders. Study 2, a year-long field study, took advantage of pre-existing differences in the proportion of women occupying leadership positions (e.g., female professors) in two naturally occurring environments—a women’s college and a coeducational college. Together, these studies investigated: (a) whether exposure to women in leadership positions can temporarily undermine women’s automatic gender stereotypic beliefs, and (b) whether this effect is mediated by the frequency with which female leaders are encountered. Results revealed first that when women were in social contexts that exposed them to female leaders, they were less likely to express automatic stereotypic beliefs about their ingroup (Studies 1 and 2). Second, Study 2 showed that the long-term effect of social environments (women’s college vs. coed college) on automatic gender stereotyping was mediated by the frequency of exposure to women leaders (i.e., female faculty). Third, some academic environments (e.g., classes in male-dominated disciplines like science and math) produced an increase in automatic stereotypic beliefs among students at the coed college but not at the women’s college—importantly, this effect was mediated by the sex of the course instructors. Together, these findings underscore the power of local environments in shaping women’s nonconscious beliefs about their ingroup.