A postsynaptic density (PSD) fraction, including some adherent subsynaptic web material, has been isolated from dog cerebral cortex by a short-procedure modification of methods of Davis and Bloom (21, 22) and Cotman and Taylor (20), using Triton X-100. The fraction has been visualized by thin-section, replica, and negative (phosphotungstic acid) staining electron microscopy and its proteins separated by high-resoltuion SDS gel electrophoresis. Morphologically, the preparation seems to be quite pure, with very little membrane contamination. The density is composed of protein, no nuclei acids, and very little phospholipids being detectable. The fraction had no ATPase or GTPase activity, but it did have a very small amount of cytochrome c oxidase activity (of a specific activity less than 0.5 percent that of a mitochondrial fraction) and a small amount of 5'- nucleotidase activity (of a specific activity between 6 and 7 percent that of a synaptic membrane fraction). Electron micrographs reveal cup-shaped structures approximately 400nm long and approximately 40nm wide, made up of apparent particles 13-28nm in diameter. However, en face views, and particularly micrographs of replicas and PTA-stained preparations, reveal a disk-shaped structure, outside diameter approximately 400 nm, in which filaments are seen to extend from the central part of the density. High resolution gel electrophoresis studies indicated some 15 major proteins and perhaps 10 or more minor ones; the predominant protein had a mol wt of 51,000, followed by ones at 45,000, 40,000, 31,000, 26,000, and several at 100,000. A comparison by gel electrophoresis of density fraction proteins with those of a lysed synaptosomal membrane fraction containing some adherent densities indicated some comigrating proteins, but the major membrane fraction protein, mol wt 52,000, was not found in the density fraction. Antibodies raised against the density fraction reacted with a preparation of solubilized synaptic membrane proteins. By both these criteria, it was considered that the density and the synaptic membrane have some proteins in common. By separately mixing (125)I-labeled myelin, synaptic vesicle, and mitochondrial fraction proteins with synaptosomes, and then isolating the density fraction from the mixture, it was concluded that a major 26,000 mol wt density fraction protein was common to both mitochondria and density, that none of the proteins of the density were contaminants from the mitochondrial fraction, that a minor approximately 150,000 band was a contaminant from the synaptic vesicle fraction, and that the moderately staining PSD fraction protein of 17,000 mol wt band was the result of contamination by the major basic protein of myelin. On the basis of the marker enzymatic assays and the mixing experiments, it is considered that the density fraction is moderately pure biochemically, and that its protein composition, aside from a few exceptions noted above, reflects its in situ character.