The fetal origins theory of adult disease suggests that term infants who are small for their gestational age have an increased susceptibility to chronic disease in adulthood as a consequence of physiologic adaptations to undernutrition during fetal life. Consistent evidence for an influence of women's dietary composition during pregnancy on growth of their babies is lacking, despite robust effects in animal experiments. We undertook a prospective observational study of 557 women aged 18-41 y, living in Adelaide, South Australia. Diet was assessed in early and late pregnancy using an FFQ. In early pregnancy, medians for energy intake, the proportion of energy derived from protein and from carbohydrate were 9.0 MJ, 17 and 48%, respectively. In late pregnancy the corresponding medians were 9.2 MJ, 16 and 49%. In early pregnancy, the percentage of energy derived from protein was positively associated with birth weight (P = 0.02) and placental weight (P = 0.07), independently of energy intake and weight gain during pregnancy, and after adjustment for potential confounders, including maternal age, parity, and smoking. Effects were stronger among women (n = 429) who had reliable data, based on prespecified criteria including the plausibility of dietary data when referenced against estimated energy expenditure. In addition, for this subgroup, the percentage of energy from carbohydrate in early and late pregnancy was negatively associated with ponderal index of the baby, and a specific effect of protein from dairy sources was identified. These data support the proposition that maternal dietary composition has an effect on fetal growth. Maternal diet in Western societies may therefore be important for the long-term health of the child.