Our understanding of spider reproductive biology is hampered by the vast anatomical diversity and difficulties associated with its study. Although authors agree on the two general types of female spider genitalia, haplogyne (plesiomorphic) and entelegyne (apomorphic), our understanding of variation within each group mostly concerns the external genital part, while the internal connections with the reproductive duct are largely unknown. Conventionally and simplistically, the spermathecae of haplogynes have simple two-way ducts, and those of entelegynes have separate copulatory and fertilization ducts for sperm to be transferred in and out of spermathecae, respectively. Sperm is discharged from the spermathecae directly into the uterus externus (a distal extension of the oviduct), which, commonly thought as homologous in both groups, is the purported location of internal fertilization in spiders. However, the structural evolution from haplo- to entelegyny remains unresolved, and thus the precise fertilization site in entelegynes is ambiguous. We aim to clarify this anatomical problem through a widely comparative morphological study of internal female genital system in entelegynes. Our survey of 147 epigyna (121 examined species in 97 genera, 34 families) surprisingly finds no direct connection between the fertilization ducts and the uterus externus, which, based on the homology with basal-most spider lineages, is a dead-end caecum in entelegynes. Instead, fertilization ducts usually connect with a secondary uterus externus, a novel feature taking over the functional role of the plesiomorphic uterus externus. We hypothesize that the transition from haplo- to entelegyny entailed not only the emergence of the two separate duct systems (copulatory, fertilization), but also involved substantial morphological changes in the distal part of the oviduct. Thus, the common oviduct may have shifted its distal connection from the uterus externus to the secondary uterus externus, perhaps facilitating discharge of larger eggs. Our findings suggest that the conventional model of entelegyne reproduction needs redefinition.