Plasma progesterone, 17-hydroxyprogesterone, 20alpha hydroxypregn-4-en-3-one levels were determined twice weekly up to 16 weeks gestation, where possible, in a twin pregnancy, in two patients who aborted spontaneously and in three patients who were treated with 'progesterone supplements' because of abnormal vaginal cytology. There was no correlation between vaginal smears and the plasms hormone levels and there was no evidence to suggest that progesterone supplements influenced clinical outcome. Compared with normal mean values the only difference was a significantly rise in progesterone and 20alpha hydroxypregn-4-en-3-one levels in the twin pregnancy after the 12th week and a precipitate fall in all hormone levels just prior to abortion. Plasma hormone levels could not be used to predict outcome.