This thesis presents the design and evaluation of a neurally-controlled ankle-foot prosthesis optimized to enhance rock climbing ability in persons with transtibial amputation. The bionic rock climbing prosthesis restores biologic performance of the ankle-foot complex. The user volitionally controls the positions of both the prosthetic ankle and subtalar joints via input from electromyography surface electrodes worn on the residual limb. We hypothesize that a climbing specific robotic ankle-foot prosthesis will result in more biological emulation than a passive prosthesis. Specifically, we hypothesize that joint angles of the hip, knee, ankle, and subtalar of a person with transtibial amputation while rock climbing are are more similar to the joint angles of a height-, weight-, and ability-matched control subject with intact limbs, compared to climbing with a passive prosthesis. To test the hypothesis, a powered, 2-degree-offreedom, neurally controlled prosthesis is built that comprises a pair of non-backdrivable linear actuators providing 16 degrees of dorsiflexion, 18 degrees of plantar flexion, and 20 degrees each of inversion and eversion. The prosthesis operates at a bandwidth and range of motion matching biological free-space motion of the ankle and subtalar joint. Climbing performance is evaluated by measuring joint angles and muscle activity during rock climbing with the robotic prosthesis and a traditional passive prosthesis, and comparing the kinematic data to that of a subject with intact biological limbs. We find that the bionic prosthesis brings the ankle and subtalar joint angles of the subject to more similar angles than the control subjects with intact biological limbs, compared to a standard passive prosthesis. These results indicate that a lightweight, actuated, 2-degree-of-freedom neurally-controlled robotic ankle-foot prosthesis restores biological function to the user during an extremely technical sport.