Neuroblastoma (NBL), one of the most common childhood solid tumors, has a distinct nature in different prognostic subgroups: NBL in patients under 1 year of age usually regresses spontaneously, whereas that in patients over 1 year of age often grows aggressively and eventually kills the patient. To understand the molecular mechanism of biology and tumorigenesis of NBL, we decided to perform a comprehensive approach to unveil the gene expression profiles among the NBL subsets. We constructed the subset-specific oligo-capping cDNA libraries from the primary NBL tissues with favorable (F: stage 1, high expression of TrkA and a single copy of MYCN) and unfavorable (UF: stage 3 or 4, decreased expression of TrkA and MYCN amplification) characteristics and randomly cloned 4654 cDNAs. Among 4243 cDNAs sequenced successfully, 1799 (42.4%) were the genes with unknown function. Excluding the housekeeping genes, an expression profile of each subset was extremely different. To determine the genes expressed differentially between F and UF subsets, we performed semiquantitative reverse transcriptase (RT)–PCR for each of the 1842 independent genes using RNA obtained from 16 F and 16 UF NBLs as template. This revealed that 278 genes were highly expressed in the F subset as compared to the UF one, while, surprisingly, only 27 genes were expressed at higher levels in the UF rather than the F subset. These differentially expressed genes included 194 genes with unknown function. Many of the genes expressed at high levels in the F subset were related to catecholamine biosynthesis, small GTPases, synapse formation, synaptic vesicle transport, and transcription factors regulating differentiation of the neural crest-derived cells. On the other hand, the genes expressed at high levels in the UF subset included transcription factors and/or receptors that might regulate neuronal growth and differentiation. The chromosomal mapping of those genes showed some clusters. Thus, our mass-identification and characterization of the differentially expressed genes between the subsets may become a powerful tool for finding the important genes of NBL as well as developing new diagnostic and therapeutic strategies against aggressive NBL.