Vaccination using cytotoxic T-lymphocyte (CTL) epitopes has become a widely used immunization strategy, especially because their structure makes them an attractive alternative to the delivery of whole proteins as immunogens. Nonetheless, their use is limited, in particular because of their specificity, being recognized only by cognate HLA alleles. The potential for immunizing a substantial portion of an ethnically diverse population using a modest number of peptides has been aided by the identification of HLA supertypes. However, the derivation of epitopes is often guided by methods that do not guarantee cross-reactivity, so we consider the feasibility of providing vaccine coverage to a multi-ethnic population under different assumptions. In particular, two large datasets are used to estimate the number of peptides needed to provide > or =90% group-specific coverage of a multiethnic population, when specificity is assumed to be either to a single serologic or molecular type. These assumptions are evaluated utilizing a clinically important epitope repertoire derived from two human cytomegalovirus proteins, and data on the in vitro memory response elicited by these peptides is presented. In summary, our combined theoretical and empiric studies suggest that 90% coverage of some ethnic groups is attainable with 11 uniquely defined HLA-restricted CTL epitopes. The derivation of four or more additional CTL epitopes is needed to attain 90% coverage of Blacks or Asians, the minimally covered groups. Ninety percent coverage of all major ethnic groups in a multi-ethnic population appears feasible without relying on cross-reactivity, but may require two to three times more CTL epitopes than estimated for serologic data, homogenous populations, or HLA alleles grouped as supertypes.