OBJECTIVE: To investigate food sources and intakes of iron, and dietary factors associated with serum ferritin levels in 6-24-month-old children. DESIGN: A cross-sectional survey employing proportionate cluster sampling was conducted in 1998/1999. Dietary intakes were assessed using a non-consecutive 3-day weighed food record. Serum ferritin and C-reactive protein were analysed from non-fasting venepuncture blood samples and general sociodemographic data were collected. SETTING: Cities of Christchurch, Dunedin and Invercargill, New Zealand. SUBJECTS: Randomly selected healthy 6-24-month-old non-breast-feeding children (n = 226). RESULTS: Total iron intakes (+/- standard deviation (SD)) among non-breast-feeding infants (< 12 months old; n = 42) and toddlers (> or = 12 months old; n = 184) were 8.4 +/- 2.9 mg day(-1) and 5.0 +/- 2.5 mg day(-1), respectively. Fifteen per cent of infants and 66% of toddlers were at risk of inadequate iron intakes. Main sources of dietary iron were infant formula (60%) for infants and cereals (31%) for toddlers. Meat contributed on average 2% and 10% of dietary iron in the infant and toddler diets, respectively. Dietary factors positively associated with serum ferritin were intakes of iron and vitamin C, whereas intakes of calcium and dietary fibre were negatively associated. For each 1% increase in percentage of energy from iron-fortified formula concomitant with a 1% decrease from dairy products, there was a 4.2% increased odds of replete iron stores (ferritin > or = 20 microg l(-1)). CONCLUSIONS: Toddlers were at higher risk of sub-optimal iron intakes than infants. Results suggest that a diet high in bioavailable iron is important for optimising the iron stores of young children in New Zealand.