The natural history of neuroectodermal tumors is still debated as far as prognostic factors are concerned; the same uncertainty applies to the optimal radiotherapy schedule and even more to the presumptive additive effect of chemotherapy. The rarity of these tumors and the heterogeneity of management make interpretation of literature data also more difficult. We evaluated clinical course in a cohort of 39 patients, including 31 with medulloblastoma (MB) and 8 with primitive neuroectodermal tumors (PNET). All patients were treated with radiotherapy, a standardized chemotherapy protocol including PCV scheme, and a second-line chemotherapy with cisplatin and etoposide (VP16) at recurrence. In 27 patients, intrathecal chemotherapy was also delivered. Median follow-up was 10.8 years. Overall, PNET had a worse outcome as compared to MB: median survival times were 42.8 vs. 92.6 months, respectively (p = 0.05). At 5 years, 45% of MB patients are alive. No significant difference in disease-free period was found between patients of different age, desmoplastic variant, tumor localization, or extent of surgery. Patients considered to be "high risk" had a significantly shorter disease-free period as compared with low-risk patients (27 vs. 54.7 months, p = 0.04). Systemic or intrathecal chemotherapy did not influence progression-free survival (PFS). However, in the majority of chemotherapy-treated patients, a low-dose craniospinal radiotherapy was also delivered. This combination of treatments may have avoided the expected increased percentage of failure. Moreover, more than half of recurrent patients had a partial response to chemotherapy that extended survival for approximately 3 years. Repeated surgery and chemotherapy at recurrence favorably influenced survival time.