40 consecutive pregnancies following treatment of infertile women with gonadotrophins were investigated. The 24-hourly excretion of oestriol, pregnanediol and chorionic gonadotrophin was measured at weekly intervals and correlated with the clinical findings. 23 pregnancies resuited in single babies, 6 in twins, 1 in triplets and 10 in abortion. Twelve single pregnancies were clinically normal and were used as a control group to estimate the mean and 95% confidence limits of the 24-hour excretions, and the mean and 95% confidence limits of the weekly rates of increase of each hormone excreted. These figures were used to determine if threatened abortion, abortion, pre-eclamptic toxaemia or multiple pregnancy could be predicted from the excretion of the hormones. Abnormally low excretion of at least 1 of the steroids occurred before clinical evidence of threatened abortion in 4 of 10 cases, abortion in 6 of 10 cases and pre-eclamptic toxaemia in 3 of 8 cases. The excretion of all steroids appeared to be raised in multiple pregnancies but the difference between the mean of these and the mean for single pregnancies was not significant, probably because of the wide scatter. The usefulness of hormone assays in pregnancy is discussed.