OBJECTIVE To determine the relation between mothers' self reported drinking habits before, during, and after pregnancy and infants' mental and motor development at 18 months of age. METHODS Follow up study of all singleton live births born to primigravidas living in Dundee and booked into antenatal clinics from 1 May 1985 to 30 April 1986. METHODS District of Dundee. METHODS 846 children aged 18 months, of whom 592 attended for assessment. METHODS Scores on Bayley scales of infant mental and motor development. RESULTS For full term children, maternal alcohol consumption was not significantly related to any adverse effect on the children's mental or motor development measures at age 18 months. After confounding factors had been controlled for, alcohol consumption before pregnancy and after pregnancy was significantly related to better motor performance and mental performance. CONCLUSIONS Pregnant women probably need not abstain from alcohol altogether as no detectable adverse relation was found between the child's mental and physical development and the mother's weekly consumption at levels in excess of 100g absolute alcohol. However, to allow for a margin of safety and taking into account the findings of an earlier phase of this study on the immediate effects on the newborn, it is recommended that pregnant women should drink no more than eight units of alcohol a week, the equivalent of about one drink a day.