Hiraizumi et al. (1973b) concluded that maternal-fetal ABO incompatibility effect upon the frequency of prenatal deaths was totally absent in two modern Japanese cities, Ohdate and Akita. However, they found a significant heterogeneity in the frequency of prenatal deaths among 16 mating types and suggested that the viability of A-bearing fetuses was higher than that of others, hence the frequency of prenatal deaths decreased with the increasing probability of producing A-bearing fetuses. Further analyses were performed in the present study and the above suggestion was confirmed. Then, the average numbers of prenatal deaths and pregnancies per couple were analyzed separately for their relationships with the probability of producing an A-bearing fetus. The results were found to be consistent with the model that the average numbers of prenatal losses per couple are approximately the same between two groups of matings, one producing and the other not producing A-bearing fetuses, but such losses occur more frequently for the A-bearing fetuses than others at the very early stage of pregnancy such that wives may not recognize such losses. Although the stages of losses are different between A-bearing and other fetuses, the net losses are nearly the same between them, so that such losses will not be reflected in the segregation frequencies among children, as shown by Hiraizumi et al. (1973a).