Bioinformatic tools and databases for glycobiology and glycomics research are playing increasingly important roles in functional studies. However, to verify hypotheses generated by computational glycomics with empirical functional assays is only an emerging field. In this study, we predicted glycan epitopes expressed by a cancer-derived mucin, MUC1, by computational glycomics. MUC1 is expressed by tumor cells with a deficiency in glycosylation. Although numerous diagnostic reagents and cancer vaccines have been designed based on abnormally glycosylated MUC1 sequences, the glycan and peptide sequences responsible for immune responses in vivo are poorly understood. The immunogenicity of synthetic MUC1 glycopeptides bearing Tn or sialyl-Tn antigens have been studied in mouse models, while authentic glyco-epitopes expressed by tumor cells remain unclear. To examine the immunogenicity of authentic cancer derived MUC1 glyco-epitopes, we expressed membrane bound forms of MUC1 tandem repeats in Jurkat, a mutant cancer cell line deficient of mucin-type core-1 β1–3 galactosyltransferase activity, and immunized mice with cancer cells expressing authentic MUC1 glyco-epitopes. Antibody responses to individual glyco-epitopes were determined by chemically synthesized candidate MUC1 glycopeptides predicted through computational glycomics. Monoclonal antibodies can be generated toward chemically synthesized glycopeptide sequences. With RPAPGS(Tn)TAPPAHG as an example, a monoclonal antibody 16A, showed 25-fold higher binding to glycosylated peptide (EC50=9.278±1.059 ng/ml) compared to its non-glycosylated form (EC50=247.3±16.29 ng/ml) as measured by ELISA experiments with plate-bound peptides. A library of monoclonal antibodies toward authentic MUC1 glycopeptide epitopes may be a valuable tool for studying glycan and peptide sequences in cancer, as well as reagents for diagnosis and therapy.