Data from a two-stage survey of the general population were used to test the proposition that the over-representation of minor affective disorders among women was restricted to those who had had children, independently of current involvement in childcare. Initial analyses supported this proposition strongly. Subsequent linear logistic analyses were largely in favour of an effect of marriage rather than of parity, but it remains possible that part of the gender differences in rates of depression arises because of the effect of parity in raising female prevalence. If substantiated, the parity effect could operate through social or biological mechanisms; although its nature is unclear, it merits further investigation.