The finding that individual cancers contain many mutant genes not present in normal tissues has prompted considerable interest in the cancer epitope landscape. To further understand such effects, we applied in silico-based epitope prediction algorithms and high throughput post hoc analysis to identify candidate tumor antigens. Analysis of 1,152 peptides containing missense mutations previously identified in breast and colorectal cancer revealed that individual cancers accumulate on average approximately 10 and approximately 7 novel and unique HLA-A*0201 epitopes, respectively, including genes implicated in the neoplastic process. These data suggest that, with appropriate manipulation of the immune system, tumor cell destruction in situ may provide a polyvalent tumor vaccine without a requirement for knowledge of the targeted antigens.