OBJECTIVE: To compare the intakes of haem and non-haem iron in iron depleted and iron replete children. DESIGN: Case-control study. SETTING: Early Childhood Centres and a long day care centre in Sydney, Australia. SUBJECTS: Children aged 12-36 months depleted in iron and controls matched for age and sex. MEAN OUTCOME MEASURES: Iron status by using plasma ferritin concentration. A three day weighed dietary intake record completed by the parents. Risk factors for iron deficiency assessed by questionnaire. RESULTS: Fifty six iron depleted and 68 iron replete children participated. The average daily intake of haem iron was significantly lower in the iron depleted group (t = 2.392, P = 0.018); there was a tendency towards a lower average daily intake of non-haem iron (t = 1.724, P = 0.086) and vitamin C (t = 1.921, P = 0.057) for iron depleted children. Low intake of haem iron (< 0.71 mg/day) was significantly associated with iron depletion with an odds ratio fo 3.0 (P = 0.005). The proportion of iron depleted children who were given whole cows' milk before 12 months of age was almost double that of iron replete children; multivariate analysis showed that both haem iron intake and age of introduction of cows' milk were independently associated with iron depletion. CONCLUSIONS: The results of this study show that, in young children in developed countries, a lower haem iron intake is a major risk factor for iron depletion; the introduction of whole cows' milk before 12 months is further confirmed as a risk factor. Parental education on nutrition should now focus on these two aspects of nutrition for infants and young children.