Pregnancy represents an immunological paradox, as underlined by the Nobel prize laureate Peter Medawar in the 1950s. This paradox is generating renewed interest with insights obtained in studies of pregnant mice and in ex vivo experiments performed with human cells and tissues. A number of molecular mechanisms have been discovered that prevent maternal placental immune effector cells located at the maternal-fetal interface from attacking fetus-derived cells. For example, maternal alloantibodies directed against paternal alloantigens expressed by the trophoblast are blocked by complement-inhibiting proteins, and maternal B cells specific for these paternal antigens are partially deleted Maternal antipaternal CD8+ cytotoxic T cells are inefficient, owing to the lack of HLA-A and HLA-B molecule expression on trophoblast target cells, together with the action of local immunosuppressive molecules, and transient tolerance of paternal alloantigens. NK cells present in the pregnant uterus and directed against fetus-derived trophoblast cells exhibit little if any cytotoxic potential. Interestingly, decidual NK cellltrophoblast interactions appear to play a physiological role in vascular uterine remodeling and in subsequent placental development. Most possible combinations of uterine NK KIR receptors and fetal HLA-C molecules expressed by the trophoblast are compatible with normal pregnancy, but the risk of severe preeclampsia appears to be far higher than normal when the mother's uterine NK cells do not express activating KIR (AA genotype) and when her fetus possesses group C2 HLA-C molecules.