Although the hippocampus is structurally quite different among reptiles, birds, and mammals, its function in spatial memory is said to be highly conserved. This is surprising, given that structural differences generally reflect functional differences. Here I review this enigma in some detail, identifying several evolutionary changes in hippocampal cytoarchitecture and connectivity. I recognize a lepidosaurid pattern of hippocampal organization (in lizards, snakes, and the tuatara Sphenodon) that differs substantially from the pattern of organization observed in the turtle/archosaur lineage, which includes crocodilians and birds. Although individual subdivisions of the hippocampus are difficult to homologize between these two patterns, both lack a clear homolog of the mammalian dentate gyrus. The strictly trilaminar organization of the ancestral amniote hippocampus was gradually lost in the lineage leading to birds, and birds expanded the system of intrahippocampal axon collaterals, relative to turtles and lizards. These expanded collateral axon branches resemble the extensive collaterals in CA3 of the mammalian hippocampus but probably evolved independently of them. Additional examples of convergent evolution between birds and mammals are the loss of direct inputs to the hippocampus from the primary olfactory cortex and the general expansion of telencephalic regions that communicate reciprocally with the hippocampus. Given this structural convergence, it seems likely that some similarities in the function of the hippocampus between birds and mammals, notably its role in the ability to remember many different locations without extensive training, likewise evolved convergently. The currently available data do not allow for a strong test of this hypothesis, but the hypothesis itself suggests some promising new research directions.