One of the fundamental goals of current strategies to develop an efficacious vaccine for AIDS is the elicitation of cytotoxic T lymphocyte (CTL) reactivities capable of recognizing cells infected with different subtypes of the human immunodeficiency virus type 1 (HIV-1). In efforts to explore new vaccine candidates by the UNAIDS/WHO Vaccine Committee, we review the most recent data concerning CTL epitopes that are conserved among the different HIV-1 subtypes. Moreover, we examine HLA allelic frequencies in several different populations, to determine those that could contribute to the goal of a cumulative phenotype frequency (CP) of at least 80%. By analyzing conserved epitopes in the context of HLA restricting alleles, we define a set of HIV-1 gene regions that may have the greatest potential to induce cross-clade reactive CTLs. The absence of well-defined correlates of immune protection that link CTL epitopes to delayed disease progression and/or prevention of infection does not permit an assignment of rank order of the most relevant component of a candidate vaccine. Thus far, most of the studies conducted in clade B-infected patients to define conserved and immunodominant epitopes indicate gag and pol gene products to be the most conserved among the HIV-1 subtypes. Moreover, anti-Pol and -Gag CTL responses appear to correlate inversely with disease progression, suggesting that they should be among the first choice of antigens to be included in a candidate vaccine construct aimed at induction of broad CTL responses. The impact of a clade B-based vaccine as a worldwide candidate capable of inducing protective immune responses can be determined only after "in vivo" studies. Meanwhile, extensive parallel studies in populations infected with non-clade B HIV-1 subtypes should define the patterns of immunodominant epitopes and HLA for comparison with the data already collected in clade B-infected subjects.