Development of knowledge about placental O2 transport (PO2) is discussed by focusing attention on the factors that determine umbilical venous PO2. In near-term pregnant sheep umbilical venous PO2 is much lower than maternal arterial PO2 and is about 20 torr lower than uterine venous PO2 in ewes who are the homozygous carriers of low O2 affinity ovine hemoglobin. Experimental evidence points to two main reasons for the low umbilical venous PO2 of sheep: (a) the uterine and umbilical circulations form an ineffective venous equilibration exchanger, and (b) a large uterine-umbilical venous PO2 gradient is required to draw O2 across a placental barrier which has a small O2 diffusing capacity relative to placental and fetal O2 demand and relative to the ineffective perfusion pattern. The latter explanation contradicts theoretical models which represent placental O2 transport as virtually 100 per cent blood flow limited. In near-term rabbits and guinea-pigs umbilical venous PO2 is also quite low, but for different reasons. In these species, the uterine and umbilical circulations form a countercurrent exchanger which allows the mother to perfuse the uterus at a very low rate. The effectiveness of countercurrent exchange is exploited to decrease the demand of pregnancy on the maternal circulation, rather than to increase the level of fetal oxygenation. There is suggestive, as yet inconclusive, evidence suggesting that in some species, notably the domestic cat, placental countercurrent exchange is combined with a low O2 affinity maternal hemoglobin and a sufficiently high uterine blood flow to produce a high level of umbilical venous PO2. The striking diversity and complexity of data about placental O2 transport demands great caution in applying comparative knowledge to the human placenta. Experimental evidence seems to indicate that the near-term human placenta is a venous equilibration exchanger, but the information which is presently available is inadequate for a firm conclusion.