To assess feasibility, acute toxicity, and efficacy of intraventricular methotrexate administered as part of the primary therapy in medulloblastoma.From 2001 to 2007, 240 patients 4 years with metastatic, 59 < 4 years with non-metastatic, 31 < 4 years with metastatic medulloblastoma).211 patients received an intraventricular access device with a subcutaneous reservoir for the application of chemotherapy. Reservoir-associated complications were documented in 57 (27%) patients, mostly due to infection (n = 32) and reservoir malfunction (n = 19), requiring removal in 39 (18%) patients. Acute neurotoxicity likely associated with intraventricular MTX was observed in 9/202 documented patients. Toxicity was usually mild, apart from one therapy-associated death due to toxic oedema followed by seizures. Of 519 treatment cycles including intraventricular methotrexate, 226 (43%) were reduced or omitted, most frequently due to the absence of an intraventricular device. Survival rates were higher in patients receiving ⩾ 75% of the scheduled intraventricular methotrexate dose compared to those receiving < 75% in both univariate and multivariate models (event-free survival (EFS), 61.5 versus 46.2%, p = 0.004; OS, 75.5% versus 60.4%, p = 0.015; hazard ratio: EFS 1.723, p = 0.016; OS 1.648, p = 0.051).Intraventricular methotrexate therapy was feasible and mostly well tolerated. Infections were the most frequent complication. A higher cumulative dose of intraventricular methotrexate was associated with better survival. Further evaluation of efficacy and late effects is warranted.