Classifications of pediatric neoplasms of the CNS are evolving from morphology-based approaches, that seek a "cell of origin" or "phase of embryogenesis," to more complicated and flexible systems that incorporate the results of immunohistochemical and molecular biology investigations. Markers of cellular proliferation are being employed as well. The development of improved classifications will depend on a reasoned application of newer methodologies to complement and, if necessary replace, the many worthy aspects of present classifications. Careful coordination among clinicians, radiologists, and pathologists is necessary to establish the clinical relevance of these findings. Application of these classifications and the techniques that make them possible will help resolve many current important issues. Among these are the validity of the PNET approach to embryonal tumors; the criteria for classification and grading of ependymomas; and the utility of subdividing cerebellar astrocytomas. Other important issues include the regional incidence and histologic prognostic features in pilocytic astrocytomas, the clinicopathologic features of well-differentiated neoplasms and hamartomatous lesions that are now being detected by magnetic resonance imaging, the utility of subclassification of medulloblastoma, and the status as an entity of the primitive polar spongioblastoma. The increasing familiarity of clinicians, radiologists, and pathologists with existing and "new" clinicopathologic entities will minimize the likelihood that well-differentiated, slowly growing, or well-circumscribed neoplasms are overgraded and treated by unnecessarily aggressive therapy. Pathologists also need to be aware that certain nonneoplastic lesions can be misconstrued as neoplasms.