Abstract: Neurodegenerative diseases of the retina affect millions. It is hoped that retinal prostheses will be a treatment for cases of blindness that are refractory to other interventions. We look at the differences between blindness, normal vision, and the type of vision that will be restored by retinal prosthesis. Specifically, retinal prostheses provide a discrete array of visual percepts (phosphenes) that represent only a portion of the visual field sampled by a video camera. The goal of the work presented here is to determine how useful this form of prosthetic vision is. First a basic experiment of the perception of contrast was performed and the results are less than expectations accounted for by the reduced spatial sampling frequency of the prosthetic vision system. Next, we showed that the performance of normally-sighted individuals using a simulation of retinal-prosthetic vision approximates that of subjects with actual implants, thus validating the use of simulation models. Finally, a very practical and careful study of the dependence of visual state on the ability of test subjects to activities of daily living is performed. The purpose of these tests is to study the impact of different levels of prosthetic vision on the quality of life of future patients receiving such implants. The outcomes of this work will serve as a guide for the development of retinal prostheses by identifying specific target levels of functional outcome for the patients. The results show that although a system with one bit of grayscale is useful if the resolution is high enough (i.e., 32x32), a system with three bits of grayscale is much more effective. However, more than three bits of grayscale were shown to not significantly improve outcomes. Although the minimum resolution needed to perform specific tasks varied, most of the fundamental tasks could be performed adequately (i.e., overall average time to completion within 5 standard deviations of that with normal vision) when the prosthetic vision array size was 16x16. A resolution of 32x32 enabled nearly all users to perform the tasks with ease (i.e., overall average time to completion within 3 standard deviations of that with normal vision). Given that the present state-of-the-art retinal prosthetic uses a 6x10 array, clearly further technological advancements are needed to improve the existing systems to a level that will be of great benefit to future patients.