Neuroblastoma (NBL) is an enigmatic tumor with heterogeneous clinical behaviors including maturation, regression, and aggressive growth. Despite recent progress in therapeutic strategies against advanced NBL, long-term outcomes still remain very poor. The prediction of cancer prognosis is one of the most urgent demands to initiate the suitable treatment of NBL. Recent papers have demonstrated that cancers can be diagnosed on the basis of gene expression profiling. We have been proceeded NBL cDNA project to collect a large number of genes expressed in NBLs, to identify the genes differentially expressed between favorable and unfavorable NBLs, and to make an NBL-proper cDNA chip for large-scale analysis of NBL tumors. Computational analysis of gene expression data in NBLs identified many prognosis-related genes and provided a classifier to predict the patient prognosis with high efficiency. Conversion of these findings into better diagnosis and treatment is now underway. Thus, molecular profiling of NBL has become a feasible tool for clinical applications.