We have conducted proteome-wide analysis of fresh surgery specimens derived from breast cancer patients, using an approach that integrates size-based intact protein fractionation, nanoscale liquid separation of peptides, electrospray ion trap mass spectrometry, and bioinformatics. Through this approach, we have acquired a large amount of peptide fragmentation spectra from size-resolved fractions of the proteomes of several breast tumors, tissue peripheral to the tumor, and samples from patients undergoing noncancer surgery. Label-free quantitation was used to generate protein abundance maps for each proteome and perform comparative analyses. The mass spectrometry data revealed distinct qualitative and quantitative patterns distinguishing the tumors from healthy tissue as well as differences between metastatic and non-metastatic human breast cancers including many established and potential novel candidate protein biomarkers. Selected proteins were evaluated by Western blotting using tumors grouped according to histological grade, size, and receptor expression but differing in nodal status. Immunohistochemical analysis of a wide panel of breast tumors was conducted to assess expression in different types of breast cancers and the cellular distribution of the candidate proteins. These experiments provided further insights and an independent validation of the data obtained by mass spectrometry and revealed the potential of this approach for establishing multimodal markers for early metastasis, therapy outcomes, prognosis, and diagnosis in the future.