The usual concern with self-reported alcohol intake is underreporting, both for general population samples (42) and pregnant women (8). However, underreporting is unlikely to be differential by genotype (under the null hypothesis of no effect on alcohol consumption) and it affects higher intake categories more than moderate and low intake. It is unlikely that under-reporting could explain the observed results, as the heavier drinkers without the A allele could underreport their consumption and declare similar intakes to the A allele carriers, introducing a bias towards the null. Quantity and frequency questions were used to derive pre-pregnancy weekly intake, shown to be generally valid (43) and reliable (44). Levels of pre-pregnancy drinking reported by ALSPAC participants included in this analysis were considerably lower compared with average alcohol intake of UK women aged 25–34 (45), a likely reflection of intentional behavioral change in preparation for the pregnancy. Figures on alcohol consumption during pregnancy were higher than those from surveys from the USA (46) and Sweden (34) showing that 12 and 30% of women reported consuming alcohol during gestation. Here the figure was
