Neurofilament proteins of mammalian axotomized peripheral axons, which regenerate effectively, resemble those of embryonic axons. However, injured centrally projecting mammalian axons, which fail to regenerate, have very different neurofilament compositions than during development. If changes in neurofilament composition after injury reflect the ability of axotomized neurons to regenerate effectively, then the neurofilaments of centrally projecting axons that can regenerate should more closely resemble those of developing axons. In this study, the neurofilament compositions of injured optic axons of the frog, Xenopus laevis, were examined, since these axons can regenerate a fully functional projection. Antibodies to phosphorylated and nonphosphorylated forms of neurofilament proteins that had been used previously to study the neurofilament composition of newly developing X. laevis optic axons were used in immunocytochemical studies to examine the return of neurofilaments to the optic nerve after an intraorbital nerve crush. Intraocularly injected wheat germ agglutinin conjugated to horseradish peroxidase was used to label the regenerating axons independently of their neurofilaments. Neurofilament immunoreactivities disappeared rapidly from crushed axons during the first week after surgery. By nine days after surgery, antibodies to nonphosphorylated forms of middle (NF-M) and low molecular weight (NF-L) neurofilament proteins and the Xenopus neuronal intermediate filament protein (XNIF) began to stain the nerve just beyond the lesion. By this time, however, growing axonal terminals had reached the optic chiasm. Antibodies to phosphorylated epitopes of NF-M began to stain axons at 15 days, just as growing axons began to arrive at the optic tectum. Nonphosphorylated high molecular weight neurofilament protein (NF-H) began to appear in axons between 18 and 21 days after surgery. Thus, the reappearance of neurofilaments during optic axon regeneration resembled the general pattern seen during development. The chief difference between development and regeneration was that neurofilament epitopes took longer to emerge during regeneration. One possibility is that cues encountered along the optic pathway influence the neurofilament composition of retinal ganglion cell axons. Then, the greater distances travelled by regenerating axons could account for the longer time taken for their neurofilament compositions to mature.