We have examined the incidence and size of infarction after occlusion of different portions of the rat middle cerebral artery (MCA) in order to define the reliability and predictability of this model of brain ischemia. We developed a neurologic examination and have correlated changes in neurologic status with the size and location of areas of infarction. The MCA was surgically occluded at different sites in six groups of normal rats. After 24 hr, rats were evaluated for the extent of neurologic deficits and graded as having severe, moderate, or no deficit using a new examination developed for this model. After rats were sacrificed the incidence of infarction was determined at histologic examination. In a subset of rats, the size of the area of infarction was measured as a percent of the area of a standard coronal section. Focal (1-2 mm) occlusion of the MCA at its origin, at the olfactory tract, or lateral to the inferior cerebral vein produced infarction in 13%, 67%, and 0% of rats, respectively (N = 38) and produced variable neurologic deficits. However, more extensive (3 or 6 mm) occlusion of the MCA beginning proximal to the olfactory tract--thus isolating lenticulostriate end-arteries from the proximal and distal supply--produced infarctions of uniform size, location, and with severe neurologic deficit (Grade 2) in 100% of rats (N = 17). Neurologic deficit correlated significantly with the size of the infarcted area (Grade 2, N = 17, 28 +/- 5% infarction; Grade 1, N = 5, 19 +/- 5%; Grade 0, N = 3, 10 +/- 2%; p less than 0.05). We have characterized precise anatomical sites of the MCA that when surgically occluded reliably produce uniform cerebral infarction in rats, and have developed a neurologic grading system that can be used to evaluate the effects of cerebral ischemia rapidly and accurately. The model will be useful for experimental assessment of new therapies for irreversible cerebral ischemia.