Comparability of population-based measures of risk and prognosis for primary central nervous system (CNS) tumors, which exhibit marked histological diversity, may be hampered by differences in detection and case ascertainment and by the lack of unequivocal and uniform histological criteria for pathological diagnoses. Furthermore, many cancer registries do not include benign tumors as defined by the International Classification of Diseases for Oncology (ICD-O), such as meningiomas, schwannomas and pituitary adenomas. We propose both a detailed and a rough uniform histological cluster scheme for data coded according to the first or second edition of the ICD-O, published in 1976 and 1990, respectively. Primary CNS tumors, typed and graded according to (modified) classification systems of Bailey and Cushing, Kernohan, Ringertz, the WHO and Daumas-Duport, are clustered as clinically relevant entities on the basis of the second edition of the WHO classification system, published in 1993. The proposed scheme identifies some of the (potential) pitfalls in the descriptive epidemiology of CNS tumors and may facilitate temporal and geographical comparability of population-based data.