Using two complementary methods in a framework that allows incorporating both environmental and household-level factors, we explore the correlates of underweight status among children. We use individual children as the units of analysis in 19 African countries, and subnational survey strata in 43 African, Asian and Latin American countries. We consider the relationship between household level demographic and health survey data, environmental factors from external geospatial data sets and two indicators of malnutrition among children aged 1–3, deviations from the international standards of weight-for-age and height-for-age. We discuss methods for data integration. In general, household determinants explain more variation than environmental factors, perhaps partly due to more error-prone measurement at the community level. Among individual children, some measures of agricultural capacity are related to lower incidence of child hunger, while among regions, measures relating to urbanness and population density show a stronger relationship. We give recommendations for further study, data collection and policy making.