Optic nerve regeneration in a lizard, Ctenophorus ornatus, is dysfunctional despite survival of most retinal ganglion cells and axon regeneration to the optic tectum. The regenerated retino-tectal projection at 6 months has crude topography but by 1 year is disordered; visually-elicited behavior is absent via the experimental eye. Here, we assess the influence of training on the outcome of optic nerve regeneration. Lizards were trained to catch prey presented within the monocular field of either eye. One optic nerve was then severed and visual stimulation resumed throughout regeneration. In the trained group, presentation was restricted to the eye undergoing optic nerve regeneration; for the untrained group, the unoperated eye was stimulated. Pupil responses returned in trained but not in untrained animals. At 1 year, trained animals oriented to and captured prey; untrained animals demonstrated minimal orienting and failed to capture prey. Regenerated retino-tectal projections were topographic in the trained but not in the untrained group as assessed by in vitro electrophysiological recording and by carbocyanine dye tracing. In vitro electrophysiological recording during application of neurotransmitter antagonists to the tectum revealed that the level of GABAergic inhibition was modest in trained animals but elevated in the untrained group; responses were mainly AMPA-mediated in both groups. We conclude that training improves the behavioral outcome of regeneration, presumably by stabilizing and refining the transient retino-tectal map and preventing a build-up of tectal inhibition. The results suggest that for successful central nerve regeneration to occur in mammals, it may be necessary to introduce training to complement procedures stimulating axon regeneration.