Abstract BACKGROUND In vitro models of pediatric brain tumors (pBT) are instrumental for both understanding the oncogenic molecular mechanisms and identifying/testing new therapeutic strategies. DNA methylation (DM) is a stable epigenetic modification recently used to classify tumors. We aim to apply DM and Copy Number Variation (CNV) profiling to characterize pBT primary cell lines and tumors. METHODS We included 36 pBT tissues from different histology (13 LGG, 9 DIPG, 9 HGG, 3 MB, and 2 Ependymomas), paired to their derived primary cultures. Cultures were established in two-dimensional (2D) or three-dimensional (3D) condition, as stem-cell or in serum-supplemented medium. For 9 cultures, both early (P2-P3) and long-term passages (>P4) were considered. Samples were analyzed for DM and CNV profiles using Illumina EPIC arrays and data compared with those of the brain tumor classifier. RESULTS At early passages all cells retained the same DM and genetic patterns of original tumors, with no differences related to 2D/3D methods or presence of serum in media. Primary cell lines analyzed at > P4 and cultured in serum diverged from the primary tumor. CONCLUSIONS DM profiles and CNV are useful tools to detect the recapitulation of pBT-derived primary cell-lines from the original tumor. Whatever subgroups tested, results suggest that in vitro models should be passaged as little as possible to retain the epigenetic and genetic alterations of the tumors and thus to be considered relevant for basic and translational biology. Ongoing experiments are aimed to determine how stable DM and CNV are in other conditions/tumor subgroups.