Reproductive immunology has moved on from the classical Medawar question of 60 years ago "why doesn't the mother reject the fetus?". Looking beyond fetal-maternal tolerance, modern reproductive immunology focuses on how the maternal immune system supports fetal growth. Maternal uterine natural killer (uNK) cells, in partnership with fetal trophoblast cells, regulate physiological vascular changes in the uterus of pregnant women and mice. These vascular changes are necessary to build the placenta and sustain fetal growth. NK cell functions in the uterus and elsewhere, including anti-viral and anti-tumour immunity mediated mostly by blood NK cells, are modulated by NK cell education, a quantifiable process that determines cellular activation thresholds. This process relies largely on interactions between self-MHC class I molecules and inhibitory NK cell receptors. By getting to know self, the maternal immune system sets up uNK cells to participate to tissue homeostasis in the womb. Placentation can be viewed as a form of natural transplantation unique in vertebrates and this raises the question of how uNK cell education or missing-self recognition affect their function and, ultimately fetal growth. Here, using combinations of MHC-sufficient and -deficient mice, we show that uNK cell education is linked to maternal and not fetal MHC, so that MHC-deficient dams produce more growth-restricted fetuses, even when the fetuses themselves express self-MHC. We also show that, while peripheral NK cells reject bone marrow cells according to the established rules of missing-self recognition, uNK cells educated by maternal MHC do not reject fetuses that miss self-MHC and these fetuses grow to their full potential. While these results are not directly applicable to clinical research, they show that NK education by maternal MHC-I is required for optimal fetal growth.