Previous experiments have shown that a substantial number of regenerating optic axons in adult frogs (Rana pipiens) are misrouted into the opposite optic nerve and retina during early stages of regeneration. This projection is maximal at 5 and 6 weeks after optic nerve crush. To further characterize this anomalous projection, small quantities of horseradish peroxidase (HRP) were injected into the right eye or right optic nerve 5 or 6 weeks after right optic nerve crush. Twenty-four hours later the animals were killed and regenerating axons anterogradely filled with HRP were reacted with the tetramethyl-benzidine method or a diaminobenzidine-CoCl2 method. Serial reconstruction tracing the course of individual axons through the optic chiasm showed that few of the axons projecting into the opposite optic nerve were collaterals of axons projecting centrally. Instead, the majority of labeled axons misdirected into the opposite nerve or contributing to an expanded projection into the ipsilateral optic tract turned out of the chiasm without branching. Many of the labeled regenerating axons had unusual trajectories within the chiasm, making abrupt turns or changing their direction of growth. Most of the axons misrouted into the opposite nerve came from portions of the chiasm nearest to the nerve of other eye. In three of eight frogs with an intact optic nerve, a small number of HRP-labeled axons were found in the left nerve after right nerve injection, but there was no indication that these axons reached the left eye. The results from this investigation suggest that the most parsimonious explanation for the chiasmal misrouting of regenerating frog optic axons is that axons are mechanically deflected into inappropriate pathways.