It has previously been thought difficult to produce hemorrhagic infarction in animals. Using the thalamic infarction model in the dog, the production of hemorrhagic infarction can be achieved consistently. In this study, the protective effect of mannitol and/or artificial blood (perfluorochemicals) on the hemorrhagic infarction was investigated. Adult mongrel dogs weighing about 10 kg each were used. Following temporal craniotomy, thalamic ischemia was produced by occluding four trunk arteries (the internal carotid, anterior cerebral, middle cerebral and posterior communicating arteries). Dogs which showed EEG changes indicative of thalamic ischemia were used in further experiments. The dogs were divided into 4 groups: (I) non-treated, (II) mannitol-treated, (III) fluorochemical-treated and (IV) mannitol and fluorochemical-treated. All dogs in each group underwent 6 hours of vascular occlusion followed by 1 hour recirculation. In order to evaluate the degree of hemorrhagic infarction, classification into 4 grades was done. Grade O: pale infarction without microscopical bleeding; Grade I: pale infarction wit microscopical bleeding; Grade II: a few sites of macroscopical petecheal bleeding; and Grade III: diffuse macroscopital petecheal bleeding. In the non-treated animals, autopsied brains showed hemorrhagic infarction in all cases. In mannitol-treated animals, some protective effect was found, especially in cases in which mannitol was administered within 60 minutes following occlusion. Hemorrhagic infarction was not suppressed in any of the fluorochemical-treated animals, but there was no hemorrhagic infarction in any of the animals treated with both mannitol and fluorochemicals. The present results are thought to indicate that these drugs administered together are effective in the treatment of hemorrhagic cerebral infarction.