Recent medical and dietetic studies report that between 12 and 30 percent of American school-aged children frequently skip breakfast. Evidence from a large body of literature spanning multiple fields suggests that proper nutrition intake in the morning has a positive association with individual health measures, personal behavior and cognitive performance. Despite these known associations, there has been little research determining the causal effect of school breakfast programs on academic outcomes of students. This paper exploits the staggered implementation of a federally funded in-classroom breakfast program, which provides meals to all students during class time, to determine the impacts of universally free school breakfasts on academic outcomes. Introducing universally free breakfasts increases math and reading test score gains by roughly 15 and 10 percent of a standard deviation on average, respectively. Gains are higher in schools where fewer students were previously participating in before school breakfasts. Moreover, these effects are slightly larger in later years of treatment, suggesting they are at least partly driven by year round benefits rather than only consumption at the time of testing. Offering universally free breakfasts increases participation--perhaps by reducing associated social stigmas--and appears to be a relatively inexpensive policy for achieving student gains in schools with below average income levels.