The effect of moderate son preference on family size is analysed using data from the 1982 Sri Lanka Contraceptive Prevalence Survey whose respondents were followed-up in the 1985 Sri Lanka Contraceptive Survey. Reported reproductive intentions on desire for additional children were not always reflected in overall contraceptive use. Complicating factors were the use of modern and traditional methods and the role of induced abortion. Longitudinal observations of the 1982-85 period relating women's individual reproductive behaviour to their intentions show no consistent behavioural difference between those with and without sons. However, at any given parity, the proportion reporting at least one additional child born during the intersurvey period was higher among those women who intended to have at least one additional child than among those who wanted to stop childbearing. This analysis suggests that son preference finds expression more through reproductive intentions than through actual fertility behaviour in the follow-up period.