Uterine flushings were obtained through the cervix (Method A) and through the wall of the uterus after hysterectomy (Method B) of ovariectomized Pony mares after s.c. injection of oestrogen for 1 week and progesterone for 2 weeks (Exp. 1). Non-pregnant and pregnant mares were flushed by Method A on Day 14 after ovulation and the flushings compared with those of non-pregnant mares injected i.v. with flunixen meglumine, a prostaglandin synthetase inhibitor, shortly before flushing (Exp. 3). Uterine flushings were also collected by Methods A and B from non-pregnant and pregnant Pony mares on Day 14. Endometrial and embryonic tissues from these mares were incubated with and without flunixen meglumine (Exp. 3). In all experiments, pregnancy had a significant effect on PGF content of uterine flushings or incubation media. Flushings from pregnant mares had reduced levels of PGF and were not influenced by collection technique (Exps 1 & 3). Non-pregnant Pony mares treated with progesterone responded to cervical stimulation (Method A) with an increase in intrauterine PGF over levels measured after hysterectomy (Method B) (Exps 1, 2 & 3). There was no effect on endometrial production of PGF in vitro by any tissue combination in a 2 h incubation in Krebs-Ringer-bicarbonate buffer but after 12 h incubation in Minimum Essential Medium endometrial PGF production was significantly higher when the endometria were from pregnant mares than from non-pregnant mares. PGF production in vitro was significantly suppressed by flunixen meglumine, by yolk sac membranes, and yolk sac and trophoblast, but not by trophoblast alone. The low intrauterine PGF levels in pregnant mares and the low in-vitro PGF production in the presence of the conceptus membranes may reflect inhibition of PGF synthesis and/or release by the embryo.