A review of the literature revealed 623 cases of optic gliomas with sufficient information to permit actuarial (life-table) analysis concerning the prognosis of this disease by the patients' age, tumor site, treatment, and presence of concomitant neurofibromatosis or extension into the hypothalamus or ventricle. All of these factors are important. The development of mathematical models led to the conclusion that these tumors, generally regarded histologically as low-grade astrocytomas, actually have a very wide but continuous range of growth rates. Some grow rapidly enough to be explained by simple exponential doubling at a constant rate, but most behave as though their growth decelerates. Decelerating growth rates make comparisons of various groups of patients difficult. No support is found for the classical hypothesis that some may be hamartomas. Inadequately treated gliomas of the optic nerve or chiasm bear about the same poor prognosis. However, tumors of the optic nerve (intracranial as well as intraorbital) have an excellent prognosis following complete surgical excision and only a slightly poorer prognosis following irradiation. About 5% of optic nerve gliomas recur in the chiasm following "complete" intraorbital excision. Patients with neurofibromatosis have about twice the recurrence rate following complete excision of an intraorbital glioma. Optic chiasmal gliomas appear to respond to irradiation with doses above 4500 rads. Patients with neurofibromatosis have about the same prognosis as patients without neurofibromatosis following irradiation of a chiasmal glioma.