Biomarkers may have prognostic and/or predictive value and have relied mainly on clinico-pathological information. Prognostic biomarkers provide information on patients' outcome irrespective of treatment, whereas predictive biomarkers provide information on the likelihood of response to a specific therapy. Biomarkers in the treatment of solid tumors were determined for many decades on protein expression by immunohistochemistry. Over the last decade, microarray-based technologies and new high-throughput sequencing methods have emerged, leading to a better understanding of tumor biology. The landmark advances in tumor genomics have highlighted specific molecular abnormalities, such as copy number alterations, mutations, and rearrangements. Several new cancer drugs target those specific molecular alterations or cell signaling pathways yielding unprecedented anti-cancer activity. Gene expression signatures have been developed in order to tailor adjuvant treatment in common tumor types. The "one size fits all" approach has been replaced by a personalized approach. The advent of massive parallel sequencing is responsible of a paradigm shift in biomarker discovery and clinical trial design on the way to what is now called "biomarker-driven cancer medicine" or "precision medicine."