Upper limb amputation disrupts most daily activities and reduces the quality of life of affected individuals. Building a suitable prosthetic limb, which can restore at least some of the lost capabilities, is a goal which has been pursued for centuries. In the last few decades, our rapidly expanding understanding of the human nervous system has unlocked impressive advances in artificial limbs. Today, commercial prosthetic hands can be controlled intuitively through voluntary muscle contractions. Nevertheless, despite leaps in the quality of modern prostheses, sensory feedback remains one of the major omissions, forcing users to rely on vision to accomplish basic tasks, such as holding a plastic cup without crushing it. Several sensory feedback strategies have recently been developed to restore tactile and proprioceptive feedback to amputees, demonstrating benefits in important areas, such as higher functional performance and increases in the sense of prosthesis ownership. Sensory feedback strategies can be distinguished based on whether the sensation they restore matches the quality (homologous feedback) or the location (somatotopic feedback) of the original sensation. Despite promising results, somatotopic tactile feedback strategies often result in unnatural sensations (e.g. electricity). Furthermore, restoration of more than a single sensory modality is rarely reported, despite being necessary to create artificial limbs capable of delivering realistic sensorimotor experiences during use. In this work, I proposed three novel and complementary strategies to improve sensory feedback restoration in upper limb prostheses. I begin by describing a non-invasive transcutaneous electrical nerve stimulation (TENS) approach aimed at restoring somatotopic tactile sensations, which is potentially applicable to all trans-radial amputees. This stimulation strategy was shown to lead to high performance during functional tasks, and compared favorably to more invasive approaches, despite a few key differences. Considering that there is no such thing as a one-size-fits-all solution for amputees, I concluded that TENS represents a viable alternative to invasive systems, especially in cases where an implant is not possible or desirable. In the second part, I proposed a sensory substitution approach to multimodal feedback, which delivered somatotopic tactile and remapped proprioceptive feedback simultaneously. This stimulation strategy relied entirely on implantable electrodes, simplifying the overall system by delivering two streams of sensory information with the same device. Using this feedback system, two amputees were able to perform interesting functional tasks, such as understanding the size and compliance of various objects, with high accuracy. Finally, I proposed a novel stimulation technique for sensory feedback designed to desynchronize induced neural activity during electrical stimulation, leading to more biomimetic patterns of activity. I discussed how this strategy could be combined with the results obtained in a recent study which I contributed to, in which we demonstrated that a model based encoding strategy resulted in more natural sensations of touch. This thesis provides evidence that advances in electrical stimulation protocols can lead to more capable prosthetic limbs. These new methods enable the delivery of multimodal, biomimetic sensory feedback and will help bridge the gap between scientific discovery and clinical translation.