Significant advances in the treatment of the medulloblastoma (MB) have been made in the last 30 years, reducing mortality by 2-fold. Further improvements in the cure rate require an increased understanding of the biology of MBs, and this will translate into refinements in their classification. Scrutiny of the cytological variation found among MBs has recently led to the concept of the anaplastic MB, which overlaps the large-cell variant and appears to share its poor prognosis. In contrast, the MB with extensive nodularity, a distinctive nodular/desmoplastic variant occurring in infants, has a better outcome than most MBs in these young patients. Building on cytogenetic studies that have drawn attention to abnormalities on chromosome 17 in over a third of MBs, research shows non-random losses on chromosomes 8, 9, 10, 11 and 16, and gains on chromosomes 1, 7 and 9. Overexpression of ErbB2 receptors and losses on chromosome 17p have been proposed as independent indicators of aggressive behaviour, while high TrkC receptor expression indicates a favourable outcome. There is a strong association between anaplastic/large-cell tumours and MYC amplification, which has previously been linked with aggressive disease, but associations between abnormalities on chromosome 17 and anaplastic/large-cell MBs and between abnormalities in the shh/PTCH pathway and the desmoplastic variant are more controversial. Classification of the MB histopathologically and according to profiles of molecular abnormalities will help both to rationalize approaches to therapy, increasing the cure rate and reducing long-term side-effects, and to suggest novel treatments.