Histopathologic assessment of 273 non-desmoplastic medulloblastomas (MBs) from children aged 3 to 16 years and entered into the SIOP/UKCCSG (International Society of Pediatric Oncology/United Kingdom Children's Cancer Study Group) PNET3 trial revealed that 47 (17%) fulfilled criteria for the recently proposed anaplastic variant. In addition, an anaplastic phenotype was focally present in all 5 (2%) large cell MBs from this series. Children with large cell MBs had the worst outcome, but there was also a significant difference between the event-free and overall survivals of children with classic MBs and those with anaplastic MBs. While objective morphometric analysis confirmed that subjective evaluation of nuclear size and variability contributed to the separation of MBs into classic, anaplastic, and large cell variants, these cytologic measures were not themselves prognostic indicators. However, anaplastic and classic MBs also possessed significantly different mitotic counts/indices, and these measures of proliferation were related to survival. Significant prognostic indicators in a multivariate survival analysis were histologic variant, metastases at presentation, and subtotal surgical excision of tumor. Our study supports the concept of an anaplastic variant among MBs, demonstrating that it has clinical utility.