Neocortical astrocytes make two types of gap junctions, intercellular ones create a functional syncytium, while reflexive gap junctions mediate autocellular coupling and serve unknown functions (Rohlmann and Wolff, 1996). Here, the question is addressed whether solitary astrocytes in vitro express connexin43 (Cx43) and establish gap junctions in the absence of intercellular contacts. In all media conditions tested, immunocytochemistry visualized Cx43-expression and gap junctions irrespective of the presence or absence of intercellular contacts. Reflexive gap junctions were associated with mechanical junctions (adherent spots and fascia adherens) connecting surface membranes and cytoskelal components, respectively. Both were characteristically located along incompletely separated borders between developing processes and/or branches. In addition, Cx43-immunoreactivity was found on some non-junctional membranes: i) intracellular vesicle clusters sited to forming processes and at the basis of filopodia; ii) the surface membrane of filopodial subpopulations usually appearing in bunches. Results suggest changes in the resumptive role of Cx43 in cultivated astrocytes: 1) Cx43 is not confined to intercellular gap junctions, it may even selectively compose reflexive ones; 2) from intracellular stores (vesicle aggregates), Cx43 may be incorporated into the surface membrane of filopodia; 3) by contacting other parts of the same cell surface (or neighboring cells), filopodia and membrane patches carrying Cx43-half channels may be essential in initial steps of gap junction formation; 4) the distribution of reflexive gap junctions is compatible with the hypothesis that autocellular coupling serves reorganization of cytoskeleton during the formation of cell processes and branches; 5) in general, gap junctions may be important for coordinating the cytoskeleton across intercellular contacts and within cells with complex shape.