CTL are by far the most important defence mechanisms against viral infections, and many attempts have been undertaken to induce protective CTL in vivo. In order to identify CTL epitopes for their possible use as peptide-vaccine candidates, HIV proteins were screened for peptide sequences which (i) fulfil the binding motif of the HLA-A2.1 molecule, and (ii) are involved in the natural immune response to HIV. From 73 nonameric peptides satisfying the binding motif, 20 peptides were synthesized and their binding to HLA-A2.1 was monitored by measuring the expression of HLA-A2.1 molecules on the cell surface of the mutant cell line T2. To evaluate the involvement in natural HIV infection, strongly binding peptides were used in cytotoxicity assays to assess their capacity to generate a peptide-specific CTL response in vitro. From 20 nonameric peptides synthesized, only five showed strong binding to HLA-A2.1. All five binding peptides had the secondary anchor residues, recently proposed by Ruppert et al. [1] to be required for binding to HLA-A2.1. The discrimination between bound and unbound peptides confirmed the importance of these secondary anchor residues which, beside the known binding motif, may dictate if a peptide can bind to HLA-A2.1 or not. In HIV- donors, no CTL activity against any of the HIV-derived peptides was detectable after a 12-day in vitro stimulation. In contrast, HIV-infected persons showed a cytotoxic response against peptide-labelled target cells, suggesting that they had developed upon HIV infection a cytotoxic immune response against the identified CTL epitopes.