Existing evidence on the impacts of parental education on child nutrition is plagued by both internal and external validity concerns. In this paper we try to address these concerns through a novel econometric analysis of 376,992 preschool children from 56 developing countries. We compare a naïve least square model to specifications that include cluster fixed effects and cohort-based educational rankings to reduce biases from omitted variables before gauging sensitivity to sub-samples and exploring potential explanations of education-nutrition linkages.