This article analyzes the magnitude and determinants of the indigenous test score gap in Bolivia and Chile (i.e., the mean difference in academic achievement between indigenous and nonindigenous children). In both countries, it finds that the gap ranges between 0.3 and 0.5 standard deviations, favoring nonindigenous children. A decomposition of achievement regressions that include classroom fixed effects suggests that 50%–70% of the difference is attributable to differences in schools and classrooms that are attended by indigenous and nonindigenous students. A smaller proportion (20%–40%) is attributable to varying endowments of family variables like parental education. The smallest proportion of the gap (10%–20%) is unexplained. Several implications for policy are discussed in light of the results.