Summary This paper presents evidence that increased earnings opportunities for girls can lower household preference for sons, as measured by the household's average reported ideal number of sons relative to ideal number of children. Using the 1995-96 Nepal Living Standards Survey, I find that reported preference for sons decreases by about one boy if expected wages for girls increase to about Rs. 20-25 per day. Anthropometric outcomes for girls, measured using weight-for-height z-scores, also improve significantly in households where there is a lower reported son preference, while there is no corresponding change in sons' health.