Central nervous system germ cell tumors (CNS GCTs) are the second most common central nervous system tumors in patients under 14 in Japan. To elucidate molecular basis of CNS GCTs, tumor samples and patients’ data were collected from the Intracranial Germ Cell Tumor Genome Analysis Consortium, a nationwide multi-center collaborative body to study CNS GCTs established in 2012. Whole exome sequencing was carried out in 41 CNS GCTs and targeted sequencing for up to 93 genes are performed in more than 200 CNS GCTs of all subtypes as well as 65 testicular GCTs. The results showed that the genes involved either in the MAPK pathway (e.g., KIT, RAS, CBL) or the PI3K/MTOR pathway (e.g., MTOR, PTEN) were collectively mutated in more than 50% of all types of GCTs. All 3 selected mutant MTOR protein had increased kinase activity and conferred enhanced cell migration and colony formation in vitro, which were completely suppressed by a dual MTOR inhibitor. A 450K array DNA methylation analysis has showed that pure germinomas but not other types of GCTs had a massive global hypomethylation. A comparison of methylation profiles between GCTs and mouse primordial germ cells (PGCs) of different developmental stage showed a striking similarity between pure germinomas and immature mouse PGC. Microdissected germinoma or teratoma components from a mixed GCT shared an identical MAPK (CBL) or PI3K (MTOR) mutation while displaying distinct methylation profiles. These findings suggest that GCTs may develop from an immature PGC that acquired MAPK/PI3K pathway mutations, and that further differentiate into distinct subtypes through epigenetic mechanisms. Thus, our combined genetic and epigenetic study suggested that CNS GCTs may be originated from PGC by MAPK/PI3K mutation and epigenetically driven to differentiate into various subtypes.