The main goal of the M4 protocol was to evaluate the efficacy of treatment excluding supratentorial radiation in patients with newly diagnosed medulloblastoma. All patients underwent surgical resection and received postoperative chemotherapy. Chemotherapy was adapted to the initial staging and prognostic factors (Group A: good-risk; Group B: poor-risk). Chemotherapy was started early after surgery, and consisted of two courses of the "eight drug in one day" regimen and two courses of high dose methotrexate. Radiotherapy was delayed until 5 (Group B) to 7 (Group A) weeks after the first course of chemotherapy. Radiotherapy was administered only to the posterior fossa and the spinal axis. Only 3/16 patients (18%) are alive and disease-free with a mean follow up of 6 years. The site of progression was supratentorial in 9 out of 13 patients and three patients had spinal and/or cerebrospinal fluid relapses. Only one patient had isolated posterior fossa relapse. The mean time to relapse was 484 days. We conclude that the chemotherapy regimens used in the M4 protocol do not allow the reduction of irradiation fields in patients with cerebellar medulloblastoma. In spite of long-term side effects on neurocognitive functions, supratentorial radiotherapy should remain a major component of medulloblastoma treatment.