Recent advances in the understanding of the mechanisms leading to tumor recognition by the immune system have shown that, at least in the case of human melanoma, the majority of cytotoxic T lymphocytes (CTL) identified in association with in vivo tumor regression after interleukin-2 therapy recognize nonmutated molecules expressed by most melanoma cells. For this reason, peptide-based or whole protein vaccination protocols against melanoma-associated antigens (MAA) are ongoing in several institutions, with the goal of inducing tumor regression by enhancing in vivo specific antitumor CTL reactivity. The rationale for the use of such vaccines is supported by: (a) preclinical evidence that vaccination with major histocompatibility complex class I restricted epitopes can enhance effectively cellular immunity, (b) evidence that potent antimelanoma CTL reactivity can be generated by repetitive in vitro stimulation of peripheral blood monocytes with MAA, and (c) evidence that the systemic administration of the same MAA can elicit antitumor CTL reactivity in vivo. As strategies are being developed for the development of sound vaccines, two basic approaches are investigated: one vaccination strategy is based on the administration of the specific amino acid sequence recognized by the CTL in association with a particular human leukocyte antigen (HLA) restriction element, and the other is based on the administration of the whole antigenic molecule, which relies on the organism's antigen-processing capabilities to render suitable the antigen for induction of HLA class I restricted CTL reactivity in vivo. Among the various factors complicating T-cell-based vaccination approaches stands the polymorphism of the HLA molecules. HLA are the most polymorphic of human genes, and because such polymorphism is clustered in the functional peptide-binding region, the binding of antigenic peptides is necessarily restricted to specific HLA alleles. This limits the interactions between CTL and antigen to specific sequences for each HLA allele. For this reason, the ability of an individual antigen to function as a T-cell immunogen in the context of different HLA allele restriction elements is an open question. It seems logical that whole-molecule vaccines have the potential advantage of broader use across patient populations. In particular, large antigenic molecules may contain multiple peptide sequences with putative binding properties for different HLA alleles, which in turn may elicit T-cell reactivity across the polymorphism of HLA. Such a concept, however, relies on the assumption that the same antigen may function with similar efficiency as an immunogen in association with different HLA alleles, independently from the epitopic sequence recognized in the various situations. This concept has been challenged recently by several practical observations and remains, in our opinion, an open question. This review will address the practical question of immunogenicity of molecules across the HLA polymorphism. We postulate that the complexity and success of the development of peptide-based vaccination strategies depend on the severity of this restriction, which is currently only incompletely studied and understood. Although no solutions are offered to the problem, emphasis is placed on the importance of this question, hopefully to stimulate the interest of other researchers, particularly in clinical settings, toward the investigation of this type of problem.