Peptides presented by HLA-A*0201 molecules on the surface of the human breast carcinoma cell line KS24.22 after IFN-gamma induction were analyzed by the "Predict-Calibrate-Detect" approach, which combines epitope prediction and high-performance liquid chromatography mass spectrometry. One of the predicted epitopes, MAGE-A1(278-286) (KVLEYVIKV), was found to be presented by HLA-A*0201, with an estimated copy number of 18 molecules/cell. HLA-A*0201 transgenic mice (HHD mice) were used to generate CTL lines that stained positive with an HLA-A*0201 tetramer folded around the KVLEYVIKV peptide and killed peptide-loaded mouse target cells expressing HLA-A*0201. IFN-gamma-treated or -nontreated HLA-A*0201 expressing HeLa cells transiently transfected with a plasmid expressing the MAGE-A1 gene stimulated in vitro cytokine production by the CTL lines. Moreover, IFN-gamma-treated KS24.22 cells, but not IFN-gamma-treated HLA-A*0201(+) MAGE-A1(-) cells or IFN-gamma-treated HLA-A*0201(-) MAGE-A1(+) cells, were killed by these CTLS: Thus, the combination of HLA epitope prediction, peptide analysis, and immunological methods is a powerful approach for the identification of tumor-associated epitopes.