Though many of the earlier workers made statements based on general impressions from injected preparations about the relative vascularity of different parts of the brain, only comparatively recently has an attempt been made to measure and express numerically the capillary development in different areas, by methods similar to those used by Krogh in studying the vascularity of muscle. Craigie has made an extensive study of the rat brain by such methods, Cobb, and Cobb and Talbot have investigated certain areas of the rabbit brain, and Dunning and Wolf certain parts of the nervous system of the cat. In the present paper a personal investigation (undertaken at the suggestion of Dr. Cobb) is reported, comparing the basal ganglia, cerebral and cerebellar cortex of the cat as regards capillary vascularity. The capillaries were demonstrated by Indian ink injection, and results expressed in millimeters of capillary per cubic millimeter of tissue. Substantiating observations were made on similar areas of monkey and human brains, but without attempting accurate measurement as in the cat material. The results indicate a striking variation in vascularity in the various areas of grey matter examined; the globus pallidus was found to be by far the least vascular area; the basal ganglia were on the whole less vascular than cerebral neocortex or cerebellar cortex, though the lateral geniculate body had as high a vascularity. To a certain extent Craigie's statement that the receptive and associative areas have a higher vascularity than the efferent areas was confirmed. Furthermore, a correlation could be made between the vascularity and the richness in ganglion cells of the various areas; there were, however, striking exceptions to this correlation—for instance, the granular layer of the cerebellum has a relatively low vascularity in view of its extremely dense cellularity. This, like Dunning and Wolf's demonstration of the relatively low vascularity of the cat's trigeminal ganglion, suggests that the metabolic activity of an area of nervous tissue (as measured by its vascularity) is not necessarily in proportion to its richness in ganglion cell bodies. The relationship, if any, of richness or poverty of capillary development to the selective localization of anoxaemic lesions remains obscure. That selective necrosis in carbon monoxide poisoning occurs in the globus pallidus, the least vascular area of grey matter so far examined, is superficially suggestive; but while the vascularity of this area is poor, so also may be its oxygen need for survival (as is suggested by its sparseness in ganglion cells), and there is no means of knowing as yet whether this or any other area has a low margin between anatomically available capillary bed and oxygen need.