The Pediatric Oncology Group (1986-1990) conducted a study in which 48 children <3 years of age with intracranial ependymomas were treated with prolonged postoperative chemotherapy (CT) and delayed RT. Thirty-one children, 0-23 months of age at diagnosis (Gp A) received 2 years of CT followed by RT; while 17 children, 24-36 months of age at diagnosis (Gp B) received CT for 1 year followed by radiation. One-year survivals were 87% (Gp A) and 94% (Gp B) and 2-year survivals were 67% (Gp A) and 82% (Gp B). In subsequent years a significant divergence in survivals according to age has been noted (p = 0.04). Five-year survivals were 25.7% (Gp A) vs. 63.3% (Gp B). The curves began to diverge 1 year following diagnosis. Other than age, the only significant prognostic factor was degree of surgical resection: 5-year survivals were 66% (total resection) vs. 25% (subtotal resection). Neither the presence of metastases, degree of anaplasia nor the degree of surgical resection varied significantly according to age at diagnosis. The most likely reason for the difference in survivals between the two age groups relates to the timing of radiation following CT, i.e., 1-year delay in children 24-36 months of age compared to a 2-year delay in children 0-23 months of age. An alternative but less likely hypothesis is that ependymomas in the younger children have a more aggressive biology. In contrast, survivals in the 24- to 36-month group are much better than previous reports in the literature suggesting that prolonged postoperative CT may allow both a delay in CRT as well as provide improved survivals. Based on these results, future treatment trials should emphasize maximal surgical resection and a delay in radiation of no more than 1 year.