The interaction between egg and sperm is the first necessary step of fertilization in all sexually reproducing organisms. A decade-long search for a protein pair mediating this event in mammals culminated in the identification of the glycosylphosphatidylinositol (GPI)-anchored glycoprotein Juno as the egg plasma membrane receptor of sperm Izumo1 [1,2]. The Juno–Izumo1 interaction was shown to be essential for fertilization since mice lacking either gene exhibit sex-specific sterility, making these proteins promising non-hormonal contraceptive targets [1,3]. No structural information is available on how gamete membranes interact at fertilization, and it is unclear how Juno — which was previously named folate receptor (FR) 4, based on sequence similarity considerations — triggers membrane adhesion by binding Izumo1. Here, we report the crystal structure of Juno and find that the overall fold is similar to that of FRα and FRβ but with significant flexibility within the area that corresponds to the rigid ligand-binding site of these bona fide folate receptors. This explains both the inability of Juno to bind vitamin B9/folic acid [1], and why mutations within the flexible region can either abolish or change the species specificity of this interaction. Furthermore, structural similarity between Juno and the cholesterol-binding Niemann-Pick disease type C1 protein (NPC1) suggests how the modified binding surface of Juno may recognize the helical structure of the amino-terminal domain of Izumo1. As Juno appears to be a mammalian innovation, our study indicates that a key evolutionary event in mammalian reproduction originated from the neofunctionalization of the vitamin B9-binding pocket of an ancestral folate receptor molecule.