Using a transorbital approach we induced the temporal occlusion and reperfusion model in 18 cats. A vascular clamp was placed on the main trunk of the left middle cerebral artery (MCA) for 1 h. Diffusion- and perfusion-weighted MR images were obtained at 1, 3, 6 and 24 h after the clip was released. The cats were killed 24 h after reperfusion, and triphenyl tetrazolium chloride (TTC) staining was performed. After the relative cerebral blood volume (rCBV), time to peak enhancement (TTP) and apparent diffusion coefficient (ADC) maps had been acquired, ROIs were drawn on (1) the area of the infarct produced, (2) the area of high signal intensity on initial diffusion-weighted magnetic resonance imaging (DWI) but normal on TTC staining, e.g., salvaged parenchyma. The ratios of these areas to the normal contralateral cortex were calculated and compared with those of the areas of the final infarct and the salvaged parenchyma. Areas of final infarct showed a temporal increase of rCBV on 3 and 6-h imaging and a final depletion on 24-h imaging. A persistent decrease of ADC value and delayed TTP were observed. Salvaged parenchyma also showed increased rCBV after reperfusion until the last imaging comparing it to the final area of infarct (P< 0.05, 24-h rCBV). The initial decrease in the ADC and delayed TTP normalized on 24-h imaging. In conclusion, rCBV of 24-h imaging was the reliable parameter to predict final infarct. A combination of serial changes on DWI and perfusion-weighted imaging (PWI) can predict ischemic penumbra and outcome.