Under what conditions do women in office leverage their power on behalf of women? Despite Latin America’s deep gender inequalities, women have democratically won the presidency more times here than in any other region in the world. Presidentas could have a major impact on these inequalities because Latin American presidents enjoy vast formal powers. Though similarly situated, Presidents Michelle Bachelet in Chile and Dilma Rousseff in Brazil dramatically diverged in their use of these prerogatives to advance pro-women change. This dissertation solves the Bachelet-Rousseff puzzle with a novel, constituency-centered theory. I argue that female politicians are most likely to deploy their power to promote change favoring women when they have (1) successfully mobilized women on the basis of gender identity (core constituency); and (2) extensively networked with elite feminists and other female politicians (personal constituency). The theory explains why female politicians overall are more likely than their male counterparts to use their power to advance pro-women change, articulates the conditions under which female politicians will make a difference and predicts the kinds of pro-women policies they are most likely to pursue. It therefore provides fresh insights on the benefits and limitations of women’s presence in political office. To illustrate and test this constituency theory, I employ newspaper archives to trace the formation of constituencies of all viable candidates in Chile’s and Brazil’s presidential races from 1999-2010. I then measure the use of legislative and delegative power to advance pro-women change by the winners of these elections with three original databases. Challenging existing theories of representation, the results of the qualitative and statistical analyses support my argument that gendered and sex-related characteristics of presidents’ core and personal constituencies largely determine the extent to which and how they use of power to promote pro-women change.