Recent technological advances have significantly improved our understanding of tumor biology by means of high-throughput mutation and transcriptome analyses. The application of genomics has revealed the mutational landscape and the specific deregulated pathways in different tumor types. At a transcriptional level, multiple gene expression signatures have been developed to identify biologically distinct subgroups of tumors. By supervised analysis, several prognostic signatures have been generated, some of them being commercially available. However, an unsupervised approach is required to discover a priori unknown molecular subtypes, the so-called intrinsic subtypes. Moreover, an integrative analysis of the molecular events associated with tumor biology has been translated into a better tumor classification. This molecular characterization confers new opportunities for therapeutic strategies in the management of cancer patients. However, the applicability of these new molecular classifications is limited because of several issues such as technological validation and cost. Further comparison with well-established clinical and pathological features is expected to accelerate clinical translation. In this review, we will focus on the data reported on molecular classification in the most common tumor types such as breast, colorectal and lung carcinoma, with special emphasis on recent data regarding tumor intrinsic subtypes. Likewise, we will review the potential applicability of these new classifications in the clinical routine.