Because sex-selective abortions are generally conducted during the second term of the pregnancy, timing of abortion can be used as an indirect way of studying sex-selection by abortion. We examined the likelihood of having a first-trimester vs. second-trimester abortion among a group of 885 married women who had an abortion in an obstetric hospital in Hanoi in 2003. In the absence of sex-selection by abortion, the number and sex of living children should not affect the timing of abortion. Results indicate that women with more children, particularly those with more daughters or without a son, were more likely to undergo a second-term abortion than a first-term abortion. We estimate that, in 2003, 2 per cent of all abortions to women with at least one living child were intended to avoid the birth of a female.