Late entry into primary school is a widespread phenomenon in developing countries. Students who enter school late are more likely to repeat grades, drop out and perform more poorly. Yet the phenomenon has received little scholarly attention, and there is a dearth of cross-national data. In this paper, we draw on data from the Multiple Indicator Cluster Survey (MICS2), a cross-national household survey conducted in developing countries. We first estimate the percentage of students entering primary school late across 38 countries in order to identify the countries in which the issue of late entry is most common. Secondly, we describe the background characteristics of students who are more likely to enter school late. We then employ multinominal logistic regression equations to predict the probability of late entry. Our findings highlight the need for policies to reduce late entry for children from disadvantaged backgrounds.