Rural schools are usually behind in terms of learning, and part of this could be related to geographical isolation. We explore this hypothesis, assessing the effect of distance between rural schools and local governments on learning in Colombia. We use spatial discontinuous regression models based on detailed administrative records from the education system and granular geographic information. Results indicate that distance to towns and Secretary of Education has significant negative effects on students’ standardized test scores. We evaluated alternative mechanisms, finding that the effect of distance is partly explained by differences in critical educational inputs, such as teachers’ education attainment and contract stability. Finally, we assess the mediating role of a program providing monetary incentives to teachers and principals in remote areas.