A placenta as we know now is a relatively new invention in mammals. Data accumulated indicates that a major cell type of the placenta is trophoblast, in which elevated expression of genes derived from various endogenous retroviruses (ERVs) as well as LTR retrotransposons is seen. However, evolutionally significance of ERV expression in placental development has not been well characterized or sorted out. In this review, we describe diversity of placental structures among mammalian species, of which morphological and cells types are far more diverse than those expected from the lines of mammalian orders. We then describe paternally expressed gene 10 (Peg10/Sirh1) and Peg11/Sirh2 as ERVs associated with ancient placenta development, followed by env-related genes such as Syncytin-1, -2, -A, -B, -Rum1, and Fematrin-1 responsible for trophoblast cells fusion, resulting in multinucleate syncytiotrophoblast formation. Because the endogenization of retroviral infections has occurred multiple times in different mammalian lineages, and some of them use similar molecules in their transcriptional activation, we speculate that ERV gene variants integrated into mammalian genomes in a locus specific manner have replaced the genes previously responsible for cell fusion. The role of cell fusion achieved by multiple successive ERV integrations is now called ''baton pass'' hypothesis, possibly resulting in increased trophoblast cell fusion, morphological diversity in placental structures, and survivability of fetuses and/or reproductive advantage in placental mammals.