A comparative study was carried out in three localities in Kenya: Kwale, Mbooni and Lwak in an attempt to identify dietary factors important in designing community based nutrition interventions. The data indicate geographical differences in age at weaning, reasons for termination of breastfeeding, the variety in the diets consumed and the nutrient intake adequacy. However, there was no difference in the duration in breastfeeding and, for children above 6 months (a period when weaning is recommended) rarely was there any food termed as "baby food" but children consumed bits and portions of the family food. In all the localities, there was no statistical difference between households with malnourished children and those without in most of the factors. Households with children admitted in the nutritional rehabilitation centres in the localities were, however, different in most of the characteristics examined. It is concluded that dietary interventions should be targeted to households and not children, mothers' care and nutrition are of paramount importance if the nutrition condition of their children is to be improved and, planning of nutritional interventions should be specific to the districts.