The pathophysiology of cerebral ischemia is well studied in small-animal models, which offer reproducibility and control of confounding variables-factors essential to hypothesis-testing. This presentation first highlights insights into the ischemic penumbra enabled by a multimodal experimental approach; second, discusses gene expression in ischemia; and third, confronts the challenges of neuroprotectant therapy. The ischemic penumbra: Transient (2-hour) middle cerebral artery suture-occlusion in anesthetized rats gives rise to highly consistent neurological and histopathological sequelae. Autoradiographic local cerebral blood flow (LCBF) studies at 2 hours of occlusion define the penumbra as a region of intermediate CBF depression (20% to 40% of control) surrounding the densely ischemic core (5% to 20% of control) and constituting one half of the entire lesion. Local glucose metabolic rate in the acute penumbra is not reduced despite the critical CBF reduction, so that the penumbral metabolism/blood flow ratio is markedly elevated. In contrast, following 1 hour of recirculation, glucose metabolism throughout the previously ischemic hemisphere has become markedly depressed, and the metabolism/flow ratio has pseudonormalized. By correlating these data with histopathology using multimodal image analysis, the probability of infarction is shown to be highly determined by the degree of antecedent CBF reduction. These animal data agree strikingly with published results in patients with acute stroke studied by positron emission tomography. This remarkable correspondence belies the assertion that data from lower species may not be relevant to human stroke. Gene expression: Perfusion gradients also determine differential patterns of gene expression in ischemia. This can be demonstrated by correlating in situ hybridization autoradiographs for gene expression with autoradiographic LCBF data and histological infarct maps derived from replicate series. In other studies, DNA microarray technology is used to screen for thousands of expressed genes. In the 2-hour middle cerebral artery occlusion model with 3-hour recirculation, we have identified 28 known ischemia-hypoxia response genes that are upregulated and 6 that are downregulated, together with 35 upregulated and 41 downregulated genes newly connected with ischemia. These findings underscore the enormous complexity of ischemic biology and suggest possible novel mechanisms for future exploration. NEUROPROTECTION: A desirable neuroprotectant would, in theory, antagonize multiple injury mechanisms. We have explored 2 such therapies of particular promise. Mild brain hypothermia (32 degrees C target temperature, for 5 hours) is highly neuroprotective even when initiated at the onset of recirculation. Another highly protective agent is human albumin, administered in doses of 1.25 to 2.5 g/kg--a therapy that reduces infarct volume in this ischemia model by 60% to 65%, markedly diminishes brain swelling, and has a therapeutic window extending to 4 hours. The careful study of rodent ischemia models can yield valuable, clinically relevant insights into the pathophysiology of ischemic stroke.