Retinal ganglion cells (RGCs) in rats were retrogradely labeled with the fluorescent tracer Fluorogold (FG) and subjected to GAP-43 and c-JUN immunocytochemistry to identify those RGCs that are capable of regenerating an axon. After optic nerve section (ONS) and simultaneous application of FG to the nerve stump (group 1 experiments), GAP-43 immunoreactive RGCs (between 2 and 21 days after ONS) always represented a subfraction of both FG-labeled (i.e., surviving) RGCs and RGCs exhibiting c-JUN. GAP-43 immunoreactive RGCs represented 22% of RGCs normally present in rat retinae and 25% of surviving RGCs at 5 days after ONS but were reduced to 2% and 1%, which is 6% and 5% of survivors at 14 and 21 days, respectively. In animals that received a peripheral nerve (PN) graft after ONS (group 2 experiments), RGCs with regenerating axons were identified by FG application to the graft at 14 and 21 days. When examined at 21 and 28 days, all FG-labeled RGCs exhibited GAP-43 immunoreactivity, and FG/GAP-43-labeled RGCs were 3% and 2% of those present in normal rat retinae. In relation to surviving RGCs GAP-43 immunoreactive RGCs represented 10% at both time points. FG-/GAP-43-labeled RGCs also exhibited c-JUN, but c-JUN immunoreactive RGCs were at both time points at least twice as numerous as FG-/GAP-43-labeled RGCs. These data suggest that regenerating axons in PN grafts derive specifically from GAP-43 reexpressing RGCs. Appearance of GAP-43 immunoreactivity may therefore identify those RGCs that are capable of axonal regeneration or sprouting.