More than one million students in the United States experience homelessness annually. Among their challenges is getting to school. This paper uses novel administrative data and a natural experiment in shelter scarcity to assess the effects of school proximity. For the average homeless K-8 student, a 10-mile longer commute leads to 6-13 percent more absences, a quarter higher probability of changing schools, and a decline in math test scores of 0.03-0.11 standard deviations. A complementary difference-in-differences design reinforces the importance of distance. The prevalence of housing instability in public schools suggests broad policy relevance.