The rotated eye paradigm has been a major experimental test of the neuronal specificity model for the development of ordered retinotectal connections in amphibians. In most studies, however, no optic fiber pathways were traced from rotated eyes and correlated with visuotectal projections. As an initial approach to this question, optic fibers from eyes rotated at different embryonic stages were traced with 3H-proline autoradiography. Three experimental series were prepared: in situ eye rotations, isochronic transplants of eyes rotated between embryos at the same stage, and heterochronic transplants of eyes rotated between embryos at different stages. Single or multiple optic fiber pathways developing from rotated eyes are identified by their sites of entry and trajectory in the brain. These include a normal chiasmatic (CH) pathway, and three aberrant pathways, identified as trigeminal (TR), diencephalic (DI), and oculomotor (OC). The latter three enter the brain ipsilaterally, some crossing contralaterally via commissural pathways. Depending on stage and type of operation, TR pathways develop in 50-100% of the animals, while CH pathways are more common after rotation at stage 21/22. The surgical procedure affects the initial trajectory of fibers from the retina, perturbs guidance cues in the surrounding orbit, and determines the patterns of optic pathways that develop. In most cases, optic fibers follow motor (oculomotor) or sensory (trigeminal) nerves, usually the first fibers encountered near the orbit by axonal pioneers exiting the retina. Evidently, optic fibers exhibit no pathway selectivity; any axon serves as a guidance cue. Tecta are innervated in about 50% of the cases, usually by fibers following abnormal trajectories from CH and OC pathways. The results suggest that the development of ordered visuotectal projections from rotated eyes is a complex process that may be independent of the trajectory of fiber arrival. Unless pathways and visuotectal maps are directly compared in each animal, however, the question remains open because we still do not know which anomalous pathways, if any, correlate with ordered projections.