A retrospective study of 86 dogs with brain tumors was undertaken. Sixty-nine dogs had histologic confirmation of tumor type, whereas the remaining 17 dogs had CT evidence of a brain tumor. All dogs had neurologic abnormalities. Seven dogs received no treatment, 38 dogs received only symptomatic treatment, and 41 dogs received some form of definitive treatment, in addition to medical management. Types of definitive treatment included surgery, cobalt-60 radiation, whole-body hyperthermia, 125I implants, and chemotherapy, alone or in combination. The factor that was most associated with survival duration was mode of therapy. Those dogs who were treated with cobalt-60 radiation, with or without other combinations of therapy, lived significantly longer than dogs who received surgery (+/- 125I implants), or dogs who received symptomatic treatment (P = 0.01 and P less than 0.001, respectively). After statistic adjustment for treatment, multiplicity of brain involvement (solitary vs. multiple) provided prognostic information with respect to survival (P = 0.001), with dogs who had a solitary site of involvement having a better prognosis. After further adjustment, initial neurologic dysfunction (mild/moderate vs. severe) showed significance as prognostic variable (P = 0.005). Both the mild and moderate groups had a more favorable prognosis compared with dogs who had severe initial neurologic impairment. The median survival time for the 86 dogs was 1.0 month (range: 1 day-42.4 mo). Median survival times of dogs receiving: 1) no therapy or only symptomatic therapy, 2) surgery (+/- 125I), or 3) cobalt-60 radiation (+/- hyperthermia, +/- surgery) were 0.2, 0.9, and 4.9 months, respectively.