Carcinoembryonic antigen (CEA) is a potential target for antigen-specific immunotherapy, as it is frequently overexpressed in human carcinomas. Moreover, an epitope derived from CEA, designated CAP1 (YLSGANLNL), has been proposed as naturally processed and presented by tumors in the human leukocyte antigen (HLA)-A*0201 context. Our aim was to fully characterize and assess the clinical relevance of the HLA-A*0201-restricted cytotoxic T lymphocyte (CTL) response against CEA. Stable and potent artificial antigen presenting cells (AAPCs) were used to evaluate T-cell response against CEA. These cells efficiently activate CTLs against tumor-derived epitopes after transduction with the antigenic peptides or full-length proteins. We found that AAPCs genetically modified to express CAP1, the agonist peptide CAP1-6D, or the whole CEA protein were not able to activate CAP1-specific CTLs from HLA-A*0201+ healthy donors or patients with colorectal carcinoma, even after multiple stimulations. In addition, we showed that a CAP1-specific T-cell clone, obtained after multiple stimulations of T cells of a HLA-A*0201+ healthy donor in vitro with autologous antigen presenting cells, recognized CEA(-) HLA-A*0201+ tumors transduced with a minigene encoding CAP1 but failed to react against HLA-A*0201+ tumor cells expressing CEA. Finally, AAPCs expressing the whole CEA protein did not induce any specific CTL response against CEA+ HLA-A*0201+ tumor cells highlighting the potential difficulty of mounting an efficacious T-cell response against this autoantigen. Altogether, our data indicate that CAP1 is not efficiently processed and presented by CEA+ tumor cells, and therefore, is not an appropriate target for T-cell-based immunotherapy.