The search continues for a favorable subgroup of patients with brain metastases in whom testing of new modalities might show a benefit in overall survival. Complete pre- and post-treatment CT evaluation of the brain was performed in 779 of the 859 patients entered into RTOG protocol 7916, a phase III study of the role of misonidazole combined with radiation therapy in the treatment of brain metastases. Pretreatment scan findings of mass effect, midline shift, massive edema, central necrosis, location of sentinel lesion, and number of lesions were correlated with length of survival for all patients as well as for each treatment group. The only characteristics that showed a statistically significant difference in survival in the overall group were the presence of < or = 3 lesions and the presence of a midline shift. The actual benefit in overall survival, however, was found to be only 3 weeks. The volume of the largest lesion prior to treatment did not correlate well with survival, nor did location of lesions. The time to response, number of responders and absolute decrease in number of lesions were similar for the four treatment arms. Patients who responded to cranial treatment had a significantly prolonged survival over those who did not respond. No CT characteristic evaluated in this study showed value as a clinically relevant prognosticator for patients with brain metastases for the overall group. Patients who fulfilled three of the four favorable clinical characteristics previously described by Diener-West (age < or = 60, KPS > or = 70, primary lesion absent or controlled and brain as sole site of metastasis), were analyzed separately. Those with < or = three lesions had a statistically significantly prolonged survival over those with four or more lesions.