Though it is not yet within our capacity to engineer the sex of a human fetus prior to conception, recent technological advances have made possible earlier and earlier detection of the sex of a fetus in utero. Where an embryo has been fertilized in vitro, detection of sex can be carried out much earlier still. These developments make it possible for couples to decide whether to continue with a pregnancy where a fetus is not of the preferred sex. To permit couples such an option would raise several ethical and policy issues. I consider ethical objections based, firstly, on there being something inherently wrong with the practice of sex selection; secondly, on there being serious untoward consequences from the practice of sex selection; and thirdly, on the wrongness of the abortions which would be required to make the practice effective. I argue that none of these is decisive against a handful of cases where sex selection might be justified, in particular, where it would help reduce the incidence of sex-linked diseases and in cases where strong economic or psychological grounds exist for endorsing parental choice over the sex of their offspring. Even granted that there are no strong ethical reasons for opposing all parental choice of sex (where this is understood to be a private decision), I claim that there are, nonetheless, powerful policy reasons for severely restricting the use of the expensive technology in the service of such private decision-making.