The results of recent model studies indicate that the variability in territorial distribution of the major cerebral arteries may be much greater than has been previously recognized. We review the literature on the cortical and intracerebral territories of the anterior, middle, and posterior cerebral arteries. Although most authors claim that these territories are relatively consistent, the results of their studies show many and considerable discrepancies. The variability described by Beevor has been neither excluded nor completely confirmed, yet somehow the concept of a relatively unchanging pattern of the peripheral cerebral vascularization has gradually settled into the literature. We discuss the considerable variability of the cerebral territories, as well as the discrepancies in investigation techniques, injection materials, and specimen conditions that could be factors producing these dissimilar results. Our study shows that there are no arguments in the literature to negate the variability of the cerebral territories. (Stroke 1991;22:1078-1084) S ince the introduction of computed tomography into neuroradiologic practice, the differential diagnosis of hemodynamic watershed infarction and thromboembolic ischemia is largely based on the location of the visualized infarct. Although a wedgeshaped infarction on empirical grounds suggests a hemodynamic pathogenesis, the primary cause is indicated as thromboembolic when the infarct is not located in one of the well-known border zones. Many templates and diagrams of the territories of the six major cerebral arteries-the anterior (ACA), middle (MCA), and posterior (PCA) cerebral arteries on both sides-provide a guide in making this diagnosis 1 " 5 and are based on the concept of a symmetrical and negligibly variable territorial distribution.This concept is also used in experimental models designed to investigate the hemodynamic behavior of the circle of Willis. In these models, the ratio of the vascular resistance of the six major cerebral arteries is an important parameter, 6 -11 yet the literature provides no accurate data on the in vivo peripheral resistance of these vessels. Thus, the ratio of resistance is estimated, based on the assumption that the peripheral resistance of the six major cerebral arteries is inversely proportional to the brain mass irrigated by each vessel.10 Because the variability of the territorial distribution is described in the literature as Received October 17, 1990; accepted April 23, 1991. relatively small in contrast to the large variability of the circle of Willis, the ratio of the peripheral resistance of the six major cerebral arteries is considered to be invariant in all models. However, the mathematical model of Hillen et al 11 and statistical analysis of the diameters of various segments of the circle of Willis 12 -13 indicate that within the circle a significant relationship exists between blood flow and vessel size. Considering the large variability of the proximal diameters of the major cerebral arteries (coefficients of v...