MR scans of 87 pediatric patients with brain stem gliomas were retrospectively reviewed to develop a new classification scheme based on MR imaging. The scheme that has been developed utilizes primarily T2-weighted images, as these most accurately show tumor extent. Tumors are characterized as to location of origin, focality, direction and extent of tumor growth, degree of brain stem enlargement, degree of exophytic growth, and presence or absence of cysts, necrosis, hemorrhage, and hydrocephalus. The use of this classification allowed identification of differences in a population of patients who were selected to be as similar as possible. This system will aid in the assessment of new protocols for treatment of brain stem tumors.