The best-known example of terminal-glycan variation is the ABO histo-blood group polymorphism in humans. We model two selective forces acting on histo-blood group antigens that may account for this polymorphism. The first is generated by the invasion of opportunistic bacterial or other pathogens that interact with the epithelial-mucosal surfaces. The bacteria adapt to the microenvironments of common host phenotypes and so create frequency-dependent selection for rarer host alleles. The second is generated by intracellular viruses, and accounts for the observed differentials between the ABO-phenotype frequencies. It is thought that viruses acquire histo-blood group structures as part of their envelope from their previous host. The presence of host antigens on the viral envelope causes differential transmission of the virus between host types owing to the asymmetric action of ABO natural antibodies. Our model simulations show that these two forces acting together can account for the major features of the ABO polymorphism in humans.