The reported English-language literature since 1969 on chemotherapy for malignant brain tumors has been collated, and reports of drugs or drug combinations used in phase II or III trials or with adjunctive therapies are reviewed. A significant fraction of the literature contained only anecdotal information, and even in the reports of more systematic trials, the criteria used to evaluate response were variable, ranging from subjective evaluation of clinical improvement to more stringent evaluations using one or more neurodiagnostic tests. A more consistent method of determining response is needed to facilitate comparison of different treatment protocols in the future, and guidelines for this are suggested. It appears that drugs of small molecular size (less than 450 daltons) and high lipid solubility have been the most effective. Single-agent chemotherapy using cell-cycle-specific drugs has, however, been disappointing. The nitrosoureas (especially BCNU) used either singly or in combination have consistently been the most effective agents in phase II and III trials. The combination of irradiation with adjunctive BCNU remains, for the present, the standard against which all other phase II and III protocols should be compared. New drugs for phase II evaluations are sorely needed. There is also need for new combination-drug and sequential-drug protocols for evaluation. The use of radiosensitizing agents with more novel radiation therapy fractionations and in combination with adjunctive drug therapies holds some promise for improving the disappointing results obtained thus far with chemotherapy for malignant brain tumors.