Student performance at university significantly influences individual decisions and future opportunities, especially in labour markets. This paper analyses the impact of local crime on student performance during higher education, with a focus on potential gender differences. Following students over their bachelor’s years, the identification strategy exploits granular local crime variation – violent and non-violent crimes – near students’ residences before sitting a final exam. We consider both spatial and temporal patterns of crime exposure by estimating a panel data model with student, exam and district-month fixed-effects to provide causal estimates. Our findings suggest that violent crimes have a negative impact on student performance, while non-violent have no significant effect. Notably, the results are mainly driven by high-ability female students, with suggestive evidence that male students in the bottom or middle parts of the grade distribution are also affected.