The identification of many tumor-associated epitopes as nonmutated "self" Ags led to the hypothesis that the induction of large numbers of self/tumor Ag-specific T cells would be prevented because of central and peripheral tolerance. We report in this study on vaccination efforts in 95 HLA-A*0201 patients at high risk for recurrence of malignant melanoma who received prolonged immunization with the "anchor-modified" synthetic peptide, gp100209-217(210M). Vaccination using this altered peptide immunogen was highly effective at inducing large numbers of self/tumor-Ag reactive T cells in virtually every patient tested, with levels as high as 42% of all CD8+ T cells assessed by tetramer analysis. From 1 to 10% of all CD8+ cells were tumor-Ag reactive in 44% of patients and levels >10% were generated in 17% of patients. These studies were substantiated using the ELISPOT assay and a bulk cytokine release assay. Although our data regarding "tumor escape" were inconclusive, some patients had growing tumors that expressed Ag and HLA-A*0201 in the presence of high levels of antitumor T cells. There was no difference in the levels of antitumor Ag-specific T cells in patients who recurred compared with those that remained disease-free. Thus, the mere presence of profoundly expanded numbers of vaccine-induced, self/tumor Ag-specific T cells cannot by themselves be used as a "surrogate marker" for vaccine efficacy. Further, the induction of even high levels of antitumor T cells may be insufficient to alter tumor progression.