International assessments of academic achievement are common. They are usually accompanied by attempts to infer the determinants of cross-country achievement gaps, but these inferences have little empirical foundation. This paper applies the Blinder-Oaxaca decomposition to the problem of explaining why primary students in Cuban schools score than Mexican students, on average, 1.3 standard deviations higher. The results suggest that no more than 30% of the difference can be explained by differing endowments of family, peer, and school variables. Of these, peer-group variables and, to a lesser extent, family variables explain the largest portion of the gap.