This paper estimates dynamic random effects models for intakes by dietary energy, protein, calcium, iron, zinc, niacin, riboflavin, thiamin, and vitamins A, C, D, and E 100 of Kenyan school children (6-9 years) within a multivariate longitudinal framework. The explanatory variables were socioeconomic and background variables, children's body mass index, and maternal education, cognitive test scores and morbidity spells. The model parameters are estimated using the maximum likelihood method controlling for unobserved between-children differences. The main finding is that while maternal education was usually not a significant predictor of dietary intakes, maternal scores on cognitive tests did strongly predict them. Moreover, the paternal cognitive scores and maternal morbidity levels were not significantly associated with the intakes, but an index of socioeconomic status and cash income was a significant predictor. The results indicate the need to consider broader measures of human development and of devising suitable educational programs for women without formal education.