Although recently survival of some medulloblastoma patients increased remarkably, it remains a serious diagnosis in others. In order to predict the postoperative prognosis in patients treated for medulloblastoma, a new staging, scoring and grading system was developed. Sixty-six patients operated on microsurgically between 1975 and 1990 at a single neurosurgical center were fully followed-up. No patient was excluded due to a poor postoperative course. Completion of commonly used radiotherapy protocols was attempted in all patients. Survival of patients was evaluated by the Kaplan-Meier method. The following 5 parameters were selected to define subgroups: patients' age, tumour location and histology, degree of resection and presence or absence of metastases. Patients older than 10 years had a better prognosis than individuals aged 10 or less (p < 0.01), patients with lateral tumours had a better prognosis than patients with midline tumours with brain stem infiltration (p < 0.05), patients with complete tumour resection had a more favourable prognosis than individuals with subtotal (p < 0.01) or partial resection (p < 0.001), patients without metastases at the time of diagnosis had a better prognosis than individuals without such evidence (p < 0.001), patients with the desmoplastic tumour variant had a better prognosis than patients with classical tumour histology (p < 0.01). According to the prognosis of a distinct subgroup, scoring points were distributed which correlated with the degree of inter-subgroup significances. The sum of a single patient's scoring points was called the total score. Based on this score, three groups of prognosis were distinguished.(ABSTRACT TRUNCATED AT 250 WORDS)