Since recent studies indicate that distinct neuropeptides and projections are associated with discrete portions of the central amygdaloid nucleus (CN), a detailed investigation of the cytoarchitecture of CN should contribute to an understanding of its organization. Qualitative and quantitative analyses of the rat CN using Nissl, Klüver-Barrera, and Golgi techniques suggests that it consists of four subdivisions. The medial subdivision (CM), which is closely associated with the stria terminalis, is narrow caudally but enlarges near the rostral pole of CN. Most neurons in CM have long dendrites that branch sparingly and have a moderate number of dendritic spines. A smaller number of CM neurons have thick dendrites with virtually no spines. Lateral to CM is the lateral subdivision (CL) which appears round in coronal sections. Neurons of CL have a very dense covering of dendritic spines and resemble medium-size spiny neurons of the striatum. Area X of Hall contains spiny neurons similar to those of CL and spine-sparse neurons that resemble medium-size spine-sparse cells of the striatum. Since area X encapsulates the lateral aspect of CL, it is termed the lateral capsular subdivision (CLC) of CN. The lateral capsular subdivision enlarges rostrally and is divided into dorsal and ventral portions by a laminar extension of the putamen. Near the rostral pole of CN a small region of tightly packed, intensely stained neurons is interposed between CL and CM. Golgi preparations reveal that this intermediate subdivision (CI) of CN contains neurons similar to those of CM. The lateral subdivision, CLC, and CM correspond, in part, to subdivisions recognized in previous Nissl studies. The intermediate subdivision has not been recognized as a distinct subdivision in previous investigations. This is the first Golgi study to recognize differences in neuronal morphology in particular subdivisions of the rat CN. The correlation of Nissl and Golgi preparations has permitted a more accurate determination of the boundaries and total extent of each subdivision than the use of Nissl techniques alone.