Although the effect of son preference on sex composition of children ever born is undetectable in national-level estimates that aggregate across all families, this article provides empirical evidence from India that son preference has two pronounced and predictable family-level effects on the sex composition of children ever born. First, data from India show that smaller families have a significantly higher proportion of sons than larger families. Second, socially and economically disadvantaged couples and couples from the northern region of India not only want but also attain a higher proportion of sons, if the effects of family size are controlled.