A total of 1440 malignant astrocytic gliomas from three Phase III trials of the National Brain Tumor Study Group were studied to document the clinical usefulness of subclassifying these lesions as either an anaplastic astrocytoma or a glioblastoma multiforme. As defined by a previous "blind" pathology review, the two groups of patients were compared as to mean age, mean duration of preoperative symptoms, and postrandomization survival. In addition, 10 histologic variables were studied in 150 patients with the anaplastic astrocytoma to establish internal correlations, and to relate specific histologic variables to patient age and postrandomization survival. There were highly significant differences in the age, duration of preoperative symptoms, and post randomization survival between the two groups. Internal correlations between histologic variables in the anaplastic astrocytoma disclosed statistically significant associations between the presence of lymphocytes and gemistocytic astrocytes. It is concluded that the subclassification of malignant gliomas into the anaplastic astrocytoma and the glioblastoma multiforme defines groups of patients that are significantly different in regard to age, duration of symptoms, and length of survival. The problems of tissue sampling are recognized, however, the assignment, by a blind pathology review, to two such different groups indicates that the classification has utility for large randomized clinical trials. The analysis of histologic variables in the anaplastic astrocytomas confirms previous suggestions that lymphocytes and gemistocytes frequently coexist in malignant gliomas, but in this study these inflammatory cells did not appear to influence survival. The study reemphasizes the association between advancing age and shorter survivals in patients with malignant gliomas.