The Human Protein Atlas project is a Swedish research initiative that uses antibody-based proteomics for large scale protein profiling in human tissues and cells. Affinity-purified antibodies are produced within the project and used for immunohistochemical staining on tissue micro arrays (TMAs) in order to map the human proteome and publish the result in a protein atlas (www.proteinatlas.org). In this thesis, TMAs were used for analysis of protein expression patterns in order to identify and explore potential biomarkers of clinical relevance. In Paper I, protein expression of SATB2 was studied in colorectal cancer. The results show that SATB2 is a sensitive and specific biomarker for colorectal cancer, staining 85% of all investigated tumors. Moreover, SATB2 in combination with CK20 showed positivity in 97% of all colorectal carcinomas and is therefore suitable as a complementary tool in clinical differential diagnostics of cancer. In Paper II, ANLN was explored as a prognostic biomarker for breast cancer. A high nuclear fraction of ANLN in breast cancer was significantly correlated to large tumor size, high histological grade, hormone receptor negative tumors, high proliferation rate and poor prognosis. Furthermore, ANLN depletion in breast cancer cell lines resulted in cell cycle arrest and cellular senescence with altered cell morphology. In Paper III, young age at breast cancer diagnosis was investigated as an independent risk factor for poor prognosis. TMAs were produced from a selection of patients from a previously defined register-based cohort. The analysis shows that young women with luminal B tumors have a 2.2-fold higher risk of dying of breast cancer compared to older women. In Paper IV, vascular expression of CD93 was explored by image analysis of the tissue-based breast cancer cohort produced in Paper III. The analysis shows that young women with breast cancer display a significantly higher CD93-positive vessel area in their tumors. High CD93-positive vessel area was significantly associated with hormone receptor negative tumors, grade, Ki-67, EGFR and a poor prognosis. In conclusion, this thesis shows that protein expression profiling using TMAs is an important tool for identifying and exploring potential novel biomarkers for cancer.