The initiation of uteroplacental circulation was investigated in early pregnancy with the following material: contact ultrasonography, 250 cases; direct-vision chorionic biopsy, 10 cases; products from voluntary interruption of pregnancy, 60 cases; and hysterectomy with the pregnancy in situ, three cases. Using all of these techniques, we were unable to demonstrate a true intervillous blood flow during the first 12 weeks. We suggest that during this period the intervillous space is bathed by an acellular fluid that could be plasma filtered by the trophoblastic shell and its endovascular cone. If this is the case, the physiology of early pregnancy must be reconsidered.