Various molecular analytical techniques have providedmore and more information which is necessary for theaccurate classification of diseases. Such informationhas enabled researchers to shift from morphology-based classification to molecular characteristics-basedclassification, also known as molecular classification.Molecular classifications of diseases can be carried outat the DNA, RNA, and protein levels. At the DNA level,for example,molecular classification canbe done basedon genetic mutation, polymorphisms, cytogeneticchanges in genomes or DNA methylation differences.Classification based on different mRNA expressionprofiles is currently a key research approach. Since itsintroduction in the 1990s, biochip technology, as a high-throughput large-scale analytic technology, has beenincreasingly applied in research on functional genomics,disease genomics, and pharmacogenomics. Rapidadvances in molecular classification have shed lighton the establishment of guidelines for the molecularclassification of tumors and personalized diagnosis andtreatment.Tumors are a heterogeneous group of systemicdiseases that develop in multiple steps/phases.Although loss of genomic stability and evading immunedestruction are two milestone events that may beobserved throughout the occurrence and developmentof tumors, the pathogenesis of tumors differs widelyamong different cancers and individuals. Dramaticadvances in basic research over the last 30 yearshave enabled practitioners to understand the pathogen-esis of malignancies at the molecular level and providedsolid bases for the prevention and treatment of tumors.The prognoses of most malignancies, advancedmalignant tumors in particular, have unfortunately notbeen fundamentally improved.Histopathology has long been the “gold standard” oftumor diagnosis and the basis for clinical treatment;however, for tumors of the same phases and stages,treatment responses and prognoses may differ vastlyamong patients even after application of the sametherapeutic strategy. In fact, malignant tumors are highlyheterogeneous at the molecular level. Tumors with thesame morphology may contain varied changes inmolecular genetics, resulting in diversity in treatmentresponses and prognoses. Individualized treatmentbased on molecular differences is thus a new directionfor tumor treatment, and molecular classification is thebasis of individualized therapy.One of the challenges in the new century is todetermine specific targeted therapies aimed at tumortypes with the same pathological origin so as tomaximize the efficacy and minimize the toxicity oftherapy. Accurate classification of tumors will obviouslybe vital toward achieving this endeavor.In our laboratory, we employed high-throughputmolecular techniques, such as gene chip and tissuemicroarray to detect 18 molecular markers by immuno-histochemistry and determined that the combinedexpression levels of seven proteins, including latencymembrane protein 1 (LMP1), CD147, caveolin-1,phospho-P70S6 kinase, survivin, matrix metalloprotei-nase 11 (MMP11), and secreted protein acidic and richin cysteine (SPARC), in tumors, together with patientsex, can distinguish patients with nasopharyngealcarcinoma (NPC) into low- and high-risk groups forprediction of NPC survival [1]. For these two groups ofpatients, the division indicates the clinical application ofindividualizedtherapy soasto preventover-treatment ofthe low-risk group and ensure appropriate treatment ofthe high-risk group. We also detected cancer stem cellsin NPC and proposed that genomic instability and DNAdamage are among the main causes of cancer stem cellformation [2,3]. In fact, the expressions of manyimportant genes, including p63, RASSF1A, VEGF,CCND1, and Hsp70, significantly differ in NPC tissue