The diagnosis of malignancies is currently based on morphology and analysis of a few molecular markers of cancer cells. On the other hand, responses to treatment and clinical outcomes of patients within existent diagnostic categories are heterogeneous. What is the reason of this situation? The inadequacy of existing diagnostic methods or existing diagnostic categories? The complete elucidation of the human genome and the development of microarray technology have heralded a new era for biological sciences and medicine, particularly in cancer (Lander, 2001). Cancer is, in essence, a genetic disease. The DNA microarray technology represents a tool capable of both, simultaneously detecting and quantifying expression of tens of thousands of genes (expression profiling) in a very short period of time (Svoboda, 2004), thus provide for us molecular portrait and new insights into the biological processes underlying complex diseases such as cancer. Therefore, DNA microarrays are used to identify new molecular markers important for diagnostic and classification of many human cancers and prediction of treatment outcome (Sorlie, 2003). The aim of this presentation is to briefly review the analyses of two model malignancies studied in recent years, to illustrate the contribution of microarray technology to "personalized medicine". I will focus on diffuse large B-cell lymphoma (DLBCL), as a model of hematological malignancies, and breast cancer, as a model of solid tumors. In both cases, microarrays studies revealed new unexpected subgroups of the disease (e.g. "Germinal Center B-cell like DLBCL" vs "Activated B-cell like DLBCL", and "Basal-like", "Luminal-like A" and "Luminal-like B" breast cancer), and this stratification was correlated with the response to therapy and patients survival. In both cases, the extrapolation of these findings to more timely efficient and cost-effective methods, such as immunohistochemistry, is recently implemented in successive steps into clinical practice (Svoboda, 2006a, Svoboda, 2006b).