The regenerating optic nerve of goldfish first reestablishes a rough retinotopic map on the tectum, then goes through an activity dependent refinement that appears to involve the elimination of inappropriate branches from early regenerated arbors. Retinotopically appropriate branches and synapses may be stabilized because the normally correlated firing of neighboring ganglion cells could cause summation of their postsynaptic responses, making them more effective. Thus, refinement of the map may be similar in several ways to associative learning. In this study, we therefore tested whether ependymin, a major protein component of the extracellular fluid that has been implicated in synaptic changes thought to be associated with learning a simple task in goldfish, may also be involved in refinement of the retinotopic map. Goldfish that had undergone unilateral optic nerve crush received intraventricular infusion of antiependymin IgG or of control IgG's beginning at 21 days postcrush. Tectal recordings from these fish at 39-56 days postcrush showed that the projection had failed to sharpen, much as in the fish with activity blocked or synchronized; the average size of the multiunit receptive fields was 31 degrees vs 11 degrees normally. The field potentials elicited from these tecta by optic nerve shock were not significantly smaller than in controls, suggesting normal levels of synaptogenesis. Control projections, identically treated but infused with either unrelated IgG or Ringer's alone regenerated normally, giving multiunit receptive fields of 12 degrees. Intact (non-regenerating) projections of the experimental fish were not rendered abnormal by the IgG treatment. Histology showed the retinas and tecta of the infused fish to be normal in appearance. The results show a specific block of sharpening by antiependymin IgG. The ependymal glia of the tectum stain positively for ependymin in normal fish, particularly the cell bodies in the ependymal layer. The tectum, particularly the ependymal layer, stains more intensely during regeneration, which appears to trigger increased synthesis of ependymins in the ependymal glia. This increase and the block of sharpening by specific antibodies to ependymin suggest a possible role for ependymin in activity dependent synaptic stabilization, possibly through its polymerization when calcium is focally depleted at active synapses.