Transection of non-tectotectal fibers in the caudal one-half of the commissure of the superior colliculus restores visual orienting to a cat previously rendered hemianopic by a large unilateral visual cortical lesion. Other observations related to this recovery phenomenon (i.e., the Sprague effect) have suggested that the caudal commissural fibers whose destruction produces the recovery 1) are contralateral afferents to the superior colliculus on the side of the cortical lesion, and 2) profoundly influence visuo-motor processing in this superior colliculus. We performed anatomical and behavioral experiments to determine which of the more than 40 contralateral collicular afferents are directly involved in the Sprague effect. To guide subsequent behavioral studies, we performed a pilot anatomical experiment in which we injected WGA-HRP unilaterally into one superior colliculus at identical retinotopic loci in each of a pair of cats. One cat was normal (control), and the other (experimental) had previously received a caudal transection of the collicular commissure. Quantitative comparison of the retrograde labeling in collicular afferents revealed that a number of mesencephalic regions contain neurons that project to the colliculus via the caudal collicular commissure. Additional collicular injections of WGA-HRP demonstrated the exact location and distribution of collicular afferent neurons within these nuclei. In the behavioral experiments, we attempted to replicate the Sprague effect by destroying the neurons giving rise to the axons in the caudal collicular commissure. Ibotenic acid lesions of these neurons were performed in cats that were hemianopic following the removal of the contralateral visual cortex. Small lesions of a "critical zone" in the rostro-lateral substantia nigra pars reticulata and possibly the overlying ventral zona incerta consistently produced a visual recovery whereas lesions of the other collicular afferents did not. Paradoxically, large nigral lesions that also included the critical zone did not result in a recovery. A conceptual framework for these findings involving striato-nigro-tecto-preoculomotor interactions is presented.