A cohort of 1163 pregnant women in two small towns in South Wales, UK, was identified and followed until the children born to them were five years of age. Growth in these children is described and a number of determinants identified. Social-class differences were very small at birth but differences in height became clear by the age of two years and in head circumference before this. In height the differences were largely accounted for by greater growth in social class I, but there was a gradient in head circumference throughout all the social classes. The social class effects gradually increased as the children became older. Parity of the mothers had a small effect on size at birth but age of the mother had no effect once parity was allowed for. Data on illnesses in the children were collected but no effect on growth could be detected. By far the most important determinant of growth which could be controlled is maternal smoking. About 40% of the women smoked, about 17% heavily (15 or more cigarettes per day) and the prevalence of smoking altered little during pregnancy. There was a graded effect of smoking on growth up to a 9% deficit in birth-weight, a 2% deficit in length at birth and a 1.5% deficit in head circumference in the babies born to the mothers who smoked most heavily (25 or more cigarettes per day) compared with non-smokers. There effects decreased with age but there were still residual effects at age five years.