A widely discussed physiological puzzle of mammalian pregnancy is the immunological paradox, which asks: why is the semi-allogenic fetus not attacked by the mother's adaptive immune system? Here, we argue that an additional, and perhaps more fundamental paradox is the question: why is embryo implantation so similar to inflammation while inflammation is also the greatest threat to the continuation of pregnancy? Equally puzzling is the question of how this arose during evolution. We call this the inflammation paradox. We argue that acute endometrial inflammation was ancestrally a natural maternal reaction to the attaching blastocyst, a situation still observed in the opossum. Eutherian implantation arose through a transformation of the acute inflammation into a process essential for implantation by causing vascular permeability and matrix reorganization as well as by suppressing the effects deleterious to the fetus. We propose that this model allows us to understand the differences between ‘good inflammation’ and ‘bad inflammation’. Further, it allows us to understand the influence of inflammation on the outcome of pregnancy and maternal health.