In 1997, the Santiago office of UNESCO implemented an assessment of student achievement in Latin America, working in collaboration with 13 Latin American ministries of education. Using a common sampling methodology and survey instruments, researchers in each country collected representative samples of data on third- and fourth-grade achievement in language and mathematics, as well as background surveys from students, parents, teachers, and principals. This article uses these data and multilevel modeling to assess the relative effectiveness of private and public schools in 10 of these countries. In particular, the article argues that many prior studies have misrepresented the private school effect by failing to control for the characteristics of student peer groups. In these studies, the achievement gap between the two sectors may partly or entirely reflect the effects of better peer group characteristics, as opposed to any substantive impact of private school practices or efficiency on the outcomes of their students. The results suggest that conditioning on a complete set of student, family, and peer characteristics explains a large portion of the observed difference in achievement between public and private schools. Across the 10 countries considered in this article, the mean private school effect is approximately zero, ranging between -0.2 and 0.2 standard deviations. The relative consistency of the findings is striking, given the diversity in the size and institutional features of the private sector across countries.