Abnormalities in the layer II neurons of human entorhinal cortex have been implicated in the pathophysiology of Alzheimer's disease and schizophrenia. The reported abnormalities are not homogeneously distributed throughout the entorhinal cortex, suggesting that layer II of entorhinal cortex may contain different subpopulations of neurons, each with a different susceptibility to pathological mechanisms. In order to investigate the possible heterogeneity of neurons in layer II of human entorhinal cortex, we first identified distinct subdivisions of human entorhinal cortex by adapting the cytoarchitectonic criteria for subdivisions of monkey entorhinal cortex described by Amaral et al. (J Comp Neurol 264:326, 1987). The morphology and regional distribution of distinct subpopulations of human layer II neurons were determined through the use of immunohistochemical techniques. Multipolar, stellate, and modified pyramidal neurons in the characteristic cell clusters or islands of layer II were immunoreactive for nonphosphorylated neurofilament proteins. The intensity of immunoreactivity for the nonphosphorylated neurofilament proteins gradually increased along the rostrocaudal axis of entorhinal cortex and was primarily due to a similar gradient in the density of labeled neurons per island. The calcium-binding protein calbindin D-28K was found in both pyramidal and nonpyramidal neurons in layers II and superficial III. The distribution of calbindin-immunoreactive neurons also depended upon the region of entorhinal cortex. In rostral entorhinal cortex, labeled neurons were scattered throughout the superficial layers, whereas in caudal entorhinal cortex, distinctive patches of small calbindin-immunoreactive neurons were found among the layer II islands. Another calcium-binding protein, parvalbumin, was present in nonpyramidal neurons in layers II and III that were distinct from those containing calbindin. The regional distribution of parvalbumin-positive neurons was very similar to that of the neurofilament immunoreactive neurons; in rostral entorhinal cortex very few parvalbumin-labeled neurons were present but their frequency gradually increased in the caudal direction. In addition, punctate parvalbumin immunoreactivity was frequently encountered in the location of the nonphosphorylated neurofilament protein-positive layer II islands. These findings demonstrate that layer II of human entorhinal cortex contains distinct subpopulations of neurons, that the relative density of each subpopulation differs across cytoarchitectonic regions, and that the patterns of distribution of these subpopulations are in some cases similar and in other cases complementary. This heterogeneity in the organization of layer II of human entorhinal cortex has important implications for the study of some neuropsychiatric disorders.