To explore the feasibility of an intraocular visual prosthesis for the restoration of sight in blind individuals the authors have performed electrical stimulation experiments in alert human subjects and in amphibian retina preparations. Subjects were legally blind from retinitis pigmentosa or age-related macular degeneration, in whom a substantial fraction of inner retinal cells is thought to remain functional. Data from 11 subjects are presented. All experienced localized visual percepts, and resolution if tested, was on the order of 1 to 2, suggesting the feasibility of at least ambulatory vision. Data from both human and animal subjects indicate that more than one cell type can be stimulated, and that selective stimulation of a particular cell type may be possible. The authors argue that a straightforward image processing and stimulus delivery system may suffice for the production of pixelized prosthetic vision, and that supervised learning of the human recipient will be more important than that of the prosthesis.