Gender-based political stereotypes pervade the media environment in the United States, and this may cause voters to automatically activate these stereotypes while evaluating politicians. In the research reported here, we investigate whether voters are able to reduce the automatic activation of unwanted stereotypes and how political sophistication influences this capacity. The current experiment uses self-reports to measure controlled stereotyping, and we develop a new eye movement metric to measure automatic stereotyping. We find that political sophisticates are more effective than novices at reducing unwanted gender-based political stereotypes. This study has two main implications for communication research. First, the results suggest that the effects of gender-based automatic stereotyping—induced by the information environment—on political judgments may not be as powerful as some of the current literature portrays them to be. Second, this study adds eye movements to the arsenal of tools available to communication scholars interested in measuring covert forms of stereotyping.