Retinitis pigmentosa affects over 1.5 million people worldwide and is a leading cause of vision loss and blindness. While retinal prostheses have shown some success in restoring basic levels of vision, only generic, “one-size-fits-all” devices are currently being implanted. In this study, we used optical coherence tomography scans of the degenerated retina from 88 patients with retinitis pigmentosa to generate models of retinal thickness and curvature for the design of customized implants. We found the average retinal thickness at the fovea to be 152.9 ± 61.3 μm, increasing to a maximum retinal thickness of 250.9 ± 57.5 μm at a nasal eccentricity of 5°. These measures could be used to assist the development of custom-made penetrating electrodes to enhance and optimize epiretinal prostheses. From the retinal thickness measurements, we determined that the optimal length of penetrating electrodes to selectively stimulate retinal ganglion cell bodies and interneuron axons in the ganglion cell layer should be 30–100 μm, and to preferentially stimulate interneurons in the inner nuclear layer, electrodes should be 100–200 μm long. Electrodes greater than 200 μm long had the potential to penetrate through the retina into the choroid, which could cause devastating complications to the eye and should be avoided. The two- and three-dimensional models of retinal thickness developed in this study can be used to design patient-specific epiretinal implants that will help with safety and to optimize the efficacy of neuronal stimulation, ensuring the best functional performance of the device for patients.