This paper exploits student-level administrative data on the population of Italian university students from 2006 to 2014 to analyze the effects of high performing (HP) male or female peers on individual academic performance, according to the gender of the student. The identification strategy is based on quasi-random variation in the exposure to HP peers across cohorts, within the same university and the same degree program. The impact of HP students, proxied by their high school final grade, is heterogeneous. We found that female HP peers have stronger positive effects than HP males, in particular with peers of the same gender. Moreover, there is evidence that the exposure to HP males can be even negative, especially for female students in competitive environments, such as the STEM fields, and for low ability students of both genders.