PURPOSE: To determine the MR and CT findings which differentiate cerebral septic embolism from thromboticinfarction. MATERIALS AND METHODS: Cerebral septic embolism was confirmed by blood culture in six patients andautopsy in two. The number, size, distribution, contrast enhancement, and hemorrhage of the lesions, as seen on MRand CT, were retrospectively analyzed, and four patients were followed up for between one week and seven months. RESULTS: In a total of eight patients, infective endocarditis (n=5) and sepsis (n=3) caused cerebral septicembolism. The number of lesions was 3~7 in six patients, over 10 in one, and innumerable in one; these varied insize from punctate to 6cm and were distributed in various aseas of the brain. Gyral infarction was noted in fivepatients; non-enhancing patchy lesions involving the basal ganglia or white matter were found in five, tinyisolated nodular or ring-enhancing small lesions involving the cortex and white matter in three, peripheralrim-enhancing large lesions in one, and numerous enhancing nodules disseminated in the cortex in one. Hemorrhagehad occurred in six. Follow-up studies in four patients showed that initial lesions had enlarged in two andregressed in two; new lesions had appeared in two. CONCLUSION: Multiple lesions of different sizes and variouspatterns which include gyral infarction, patchy or nodular lesion in the cortex, white matter or basal ganglia,and isolated small ring-like or nodular enhancement or frequent hemorrhage are findings which could be helpful inthe radiologic diagnosis of cerebral septic embolism.