INTRODUCTIONPediatric brain tumors (PBT) represent the second most common pediatric cancer, with the highest mortality rate among childhood malignancies. Improvement of PBT diagnostic accuracy is fundamental to optimize treatment strategy. OBJECTIVES: We aimed to explore the impact of DNA methylation arrays implementation in PBT clinical practice.METHODS214 PBT were analyzed by Illumina 850KEPICmethylation array. Low score and discordant cases were collegially reviewed.RESULTSCalibrated score was 0.8 or higher in 159 cases (74.3%), with pathological diagnosis confirmation in 132 cases and molecular subgroup definition in 47 of them, including cases of medulloblastoma, CNS neuroblastoma FOXR2, HGNET MN1; methylation profiling amended diagnosis in 10 cases, e.g. HGNET BCOR and anaplastic PXA, was non-contributory in 4 and misleading in 12 cases, including glioneural tumors and tumors arising in syndromic contexts. Calibrated score ranged between 0.8 and 0.3 in 37 cases (17.3%) and was below 0,3 (no match) in 18 cases (8.4%). Calibrated score below 0,8 was more frequently assigned to low grade gliomas and low grade glioneural tumors (p <0.0006). Challenging/very rare cases, e.g. intracranial AFH with EWSR1:CREM fusion and nonRELA supratentorial ependymomas, were assigned to “no match subgroup”; in syndromic patients the score tended to be lower (p=0.07); no correlation between score and age < 3-years was found (p=0.1).CONCLUSIONMethylation profiling refine on diagnostic accuracy in PBT classification. Improvements are needed in classifying low grade glioma/glioneuronal tumors and challenging/very rare PBT. In syndromic cases, there is a high rate of misleading profiles and/or low scores.