To examine the usefulness of diffusion-weighted imaging for detecting neuronal damage following ischemia, dynamic changes in diffusion-, T1- and T2-weighted images of rats subjected to 10 min of 4-vessel occlusion and of humans who had suffered 10-20 min of cardiac arrest were observed. In rats, no remarkable alteration was observed on day 1. On day 3, however, diffusion-weighted images showed high signal intensity in the hippocampal area, in which the apparent diffusion coefficient was significantly lower than that of the control (760+/-28x10(-6) mm(2)/s in control vs. 480+/-29x10(-6) mm(2)/s on day 3, P<0.0001). Histological observation revealed microvacuolation in 92+/-4% of pyramidal neurons in the CA1 region. On day 7, the hyperintensity in diffusion-weighted images had disappeared and microvacuolation had also disappeared in the CA1 region, but severely disrupted pyramidal neurons containing pyknotic nuclei had appeared in the CA1 region instead. In humans, diffusion-weighted images did not show any apparent abnormality in the cerebral cortex on the day of resuscitation. On day 3, however, diffusion-weighted images consistently showed hyperintensities in the temporal or occipital cortex, and these hyperintensities had disappeared in images obtained on days 7 and 14. From day 14, T1-weighted images showed laminar hyperintensity, suggesting laminar necrosis, along the cortex, where diffusion-weighted images showed high signal intensity on day 3. These results suggested that diffusion-weighted imaging has a potential for detection of the occurrence of microvacuolation and is useful for detecting the progression of ischemic changes in humans following global ischemia.