Growth cones are highly motile extensions of growing neuronal processes that have a well-characterized cytoskeleton, which is necessary for motility and directed growth. In culture, neurons have been observed to extend processes on a variety of substrates made of cell adhesion molecules (CAMs) or extracellular matrix molecules (ECM molecules). We have previously shown that retinal ganglion cell (RGC) growth cones exhibit characteristic morphologies dependent on the substrate on which they are grown (Payne et al., 1992: Cell Motil. Cytoskel. 21:65-73). Upon contact with a sharp border between two substrates, the growth cones display rapid changes in morphology (Burden-Gulley et al., 1995: J. Neurosci. 15:4370-4381) that may result from extensive restructuring of the cytoskeleton. In the present study, immunocytochemical methods were used to examine the distribution of three cytoskeletal elements in RGC growth cones growing on L1, N-cadherin, or laminin as well as on dishes coated with alternating lanes of these substrates. Distinct distribution patterns of f-actin, microtubules (MTs), and neurofilaments (NFs) were observed in growth cones growing on individual substrates. At border regions between two substrates, growth cones with extensive lamellipodial contact with the second substrate were observed to have f-actin and MT distribution patterns appropriate for the new substrate encountered. Contact via filopodia alone did not evoke this change. Redistribution of NFs was observed only after the majority of the growth cone had crossed onto the second substrate. These results suggest that actin and MTs, but probably not NFs, are directly influenced by CAMs and ECM molecules to produce changes in growth cone morphology. The distribution of two members of the protein tyrosine kinase family, pp60arc and p59fyn, and phosphorylated tyrosine residues was also examined. No differences were observed in the distribution patterns of the kinases and phosphorylated tyrosine residues in growth cones on any of the substrate molecules tested. In addition, the distribution patterns were unchanged in growth cones that contacted and crossed borders between two substrates. These results suggest that redistribution of pp60arc or p59fyn is not required to produce alterations in growth cone morphology induced by contact with L1, N-cadherin or laminin.