The distribution of a dopamine- and cyclic adenosine-3':5'-monophosphate (cAMP)-regulated phosphoprotein with an apparent molecular weight of 32,000 (DARPP-32) was investigated in the rat diencephalon and monkey hypothalamus by use of immunohistochemical techniques. In addition to single cells located peri- and paraventricularly in hypothalamus and thalamus in the rat, and ependymal cells, DARPP-32-immunoreactivity was found to be present in a subpopulation of ependymal tanycytes. These DARPP-32-positive tanycytes lined the walls and floor of the third ventricle, sending processes towards the arcuate nucleus, surrounding blood vessels in this nucleus, and continuing towards the median eminence, where they abutted on portal vessels. A second group of DARPP-32-positive tanycytes with cell bodies within the median eminence was also observed. Simultaneous labeling with antiserum against tyrosine hydroxylase, a presumptive marker for tuberoinfundibular dopamine neurons, revealed a close relation to DARPP-32-containing tanycytes in several anatomical locations. Thus, in the periventricular area DARPP-32-positive tanycytes ensheathed tyrosine hydroxylase-positive processes. These processes, presumably representing dopaminergic dendrites, virtually penetrated between the ependymal cells to the ventricular space and thus perhaps established direct contact with the cerebrospinal fluid. Tyrosine hydroxylase-terminals were also observed in close association with DARPP-32-immunoreactive tanycytes in the rat median eminence. However, in view of the density of DARPP-32-positive processes in the external layer of the median eminence, the DARPP-32 processes may be related to a number of other types of nerve endings, including luteinizing hormone-releasing hormone, as shown in this study. The close association of DARPP-32-immunoreactive processes with tyrosine hydroxylase- and luteinizing hormone-releasing hormone-immunoreactive nerve endings in the rat was directly visualized at the ultrastructural level using triple-labeling immunocytochemistry. Both the ultrastructural analysis and immunohistochemistry at the light microscopic level, comparing the distribution of DARPP-32 and glial fibrillary acidic protein, indicated the presence of two types of glial processes in the median eminence. The electron microscopic studies also suggested the presence of both DARPP-32-positive and DARPP-32-negative glial processes in the external layer of the median eminence.(ABSTRACT TRUNCATED AT 400 WORDS)