Using novel data on students' performance on national exams administered during secondary schooling in Tanzania, we study how air pollution exposure on the day of the exam affects student performance on these exams. To uncover causal effects, we leverage plausibly exogenous changes in local wind direction in an Instrumental Variables (IV) setup. Our IV estimates imply that an increase in PM2.5 concentration by 10 µg/m3 on the day a student appears for the exam worsens their performance on the exam by 0.06 standard deviations. Our results are robust to a host of falsification checks. We also document that the effects are more pronounced for younger students, males, students appearing for exams in government schools, and those at the lower end of the achievement distribution. Further, we find that these effects could be driven by adverse effects of air pollution on exams that test fluid intelligence.