Intensive rehabilitation, including exercises with individualized difficulty, is recommended to promote motor recovery after stroke or spinal cord injury. This thesis explores how robotics and serious games can support the implementation of a consolidated set of neurorehabilitation principles. We clinically validated an adaptation algorithm that adjusts in real time exercise difficulty within a serious game implemented on a distal end-effector robot, combining cognitive and upper-limb motor rehabilitation after stroke. Two meta-analyses and one systematic review synthesized evidence on the effectiveness of serious games and self-rehabilitation after stroke, as well as on the use of self-balancing exoskeletons for gait training in individuals with high tetraplegia. The integration of these technologies, complementing conventional therapies, constitute an effective strategy in neurorehabilitation.