Although the mammalian epidermal basal cell hemidesmosome bears some superficial resemblance to one half of a desmosome at the ultrastructural level, examination of the structure of the electron-dense submembranous plaques of the hemidesmosome and desmosome reveals that they differ with respect to their overall morphology and dimensions. Based on these findings, we wondered whether components of the desmosome are present in the hemidesmosome. In order to determine this we prepared a number of stratified squamous epithelial tissues for indirect immunofluorescence using antibody preparations directed against known desmosome components including desmoplakin and certain glycoproteins. These antibody preparations do not show reaction with hemidesmosomes by indirect immunofluorescence criteria. We have also utilized bullous pemphigoid (BP) autoantibodies that have been shown to recognize hemidesmosomes in mammalian skin cells [Mutasim et al., J. Invest. Derm., 84:47-53, 1985]. Double label indirect immunofluorescence observations of neonatal mouse skin prepared using desmoplakin antibodies and BP autoantibodies reveal that hemidesmosomes that are stained by the BP autoantibodies are not recognized by the desmoplakin antibodies. We confirmed these findings at the ultrastructural level by indirect immunogold localization of desmoplakin antibodies and BP autoantibodies. Therefore, the hemidesmosome does not appear to be one half of a desmosome and may possess a very different molecular organization relative to the desmosome. We raise the possibility that the variability between the hemidesmosome and desmosome that we detect at the morphological and immunological level may reflect the functional differences of these two types of junctions.