This paper examines how the allocation of teachers across public primary schools differs between countries and the extent to which this can explain differences in educational outcomes. First, I build a new global school-level data set that comprises nearly two million schools representing public primary education in 91 countries. I document that pupil-teacher ratios (PTRs) in developed countries are low on aggregate and vary little between schools. In contrast, in developing countries aggregate PTRs are high and differences in PTRs between schools are large. Even at the local level, within second-tier administrative units, differences in PTRs between schools are substantial. While PTRs are higher in rural areas, PTR differences between schools within both urban and rural areas are much larger than differences in average PTRs between urban and rural areas. High PTRs are typically found in areas with low levels of wealth and adult literacy, and poor school infrastructure. Second, I build a model of education production to assess if complementarities between teachers, school infrastructure and household inputs can rationalize the prevailing inequalities in the relative number of teachers within developing countries. Simulations suggest that more equal teacher allocations could in fact increase, rather than decrease, aggregate learning in many poor countries. Obtaining equivalent gains through reductions in aggregate PTRs, while holding relative PTRs between schools fixed, on the other hand, would require large teacher workforce increases.