Tumor entities of the Central Nervous System (CNS) are defined by the WHO classification and range from benign neoplasms to highly malignant tumors, making histopathological classification challenging. This diverse spectrum complicates clinical decision-making and the interpretation and validation of clinical trial results. The methodology based on DNA methylation profiling of CNS tumors offers greater diagnostic precision compared to traditional morphological methods. In this study, we analyzed 16 DNA samples from medulloblastomas and ependymomas, obtained from paraffin-embedded blocks from the Pathology Division archives of HCFMUSP, and with prior histological analysis. The samples were processed using the Infinium MethylationEPIC BeadChip (Illumina®) and methylation data analysis tools in R Studio. The data were submitted to the MolecularNeuropathology.org platform for methylation based classification and CNV chart generation. The classifier was consistent with the previous diagnosis in all cases, refining the diagnosis in 54% of cases to subclass, 38% to subtype, and 8% to family. The platform also provided CNV estimates for 87% of cases, revealing crucial information on genetic alterations such as gains, losses, amplifications, and gene fusions. Despite the challenges of implementation within the Brazilian Public Health System (SUS), the results demonstrate the method's robustness and reproducibility, promising invaluable support for precise diagnoses and the development of more effective therapeutic strategies.