This paper has three main aims: to measure the clustering of children with low weight for age z-scores within families, to establish whether significant differences exist by gender in weight for age z-scores, and to demonstrate whether the presence of a mother-in-law in the household has any significant impact on the nutritional status of young children. Regression modelling is used to examine the weight for age z-scores of children under the age of four years in Maharashtra, Tamil Nadu and Uttar Pradesh using the 1992-93 Indian National Family Health Survey data. Random effects models measure the clustering of children with low weight for age z-scores in families, controlling for a number of other family factors. Our findings do not reveal significant gender differences in weight for age z-scores. Although little variation was found between family structures in the nutritional status of children, there were significant differences between families after controlling for family type. This suggests that there are differences between families that cannot be explained by a cross-sectional demographic survey. The evidence from this work suggests that nutrition programs need to adopt community nutrition interventions that aim resources at young children from families where children with low weight for age z-scores are found to cluster. However, there is a need for further inter-disciplinary research to collect data from families on behavioural factors and resource allocation in order that we might better understand why some families are more prone to having children with low weight for age z-scores. The diversity in the significant covariates between the three states in the models has shown the need for Indian nutrition programs to adopt state-specific approaches to tackling malnutrition.