While child health reflects a country’s state of well-being; Bangladesh struggles to make a comprehensive success in reducing its widespread child under-nutrition. The key interest of this paper is to understand what determines the nutritional status of rural children in Bangladesh and how these determinants compare between agricultural and non-agricultural households. Based on a sample of 1,444 children drawn from rural Bangladesh, this study estimates the extent to which measures of child under-nutrition (underweight and stunting) are associated with child, parent, household, and community level variables. The estimates of Logit regressions suggest that variables such as birth weight, parents’ health, mother’s education, and prevalence of health care facilities at village significantly reduces the probability of underweight and stunting. This paper also finds that agricultural and non-agricultural households differ in respect of what determines the children’s nutrition in these households. The policy implication of this study is that government initiatives aiming at reducing child under-nutrition in rural Bangladesh should recognise that any generic measures may not bring about the optimal result given the diversity of rural households. Thus, this paper may be useful in designing better interventions for improving health and nutrition of the rural children of Bangladesh as well as similar Southeast Asian countries.