I use a combination of blind and non-blind test scores to show that middle school teachers favor girls when they grade. This favoritism, estimated in the form of individual teacher effects, has long-term consequences: as measured by their national evaluations three years later, male students...