In this thesis, we contribute to the field of rehabilitation robotics by designing haptic-enabled tangible robot-based activities and exploring their added value for therapy and assistance. The research specifically focuses on the design and development of gamified robot-enhanced therapy and training activities for patients suffering from neurological disorders, children with neurodevelopmental difficulties and elderly. We adopted an iterative design approach to develop adaptive components tailored to the specific needs of the multiple target groups, and introduced gamification as a way to increase users' engagement with the training. The key accomplishments of our research are: (1) We designed and developed robot-mediated gamified handwriting activities by iteratively adapting, testing and integrating the system into occupational therapy environments for children with neurodevelopmental difficulties. We showed that haptic-enabled tangible robots can be a useful tool for handwriting training in multi-child therapy settings. (2) We designed and developed a novel robot-enhanced upper limb rehabilitation game, the Tangible Pacman Game, with various adaptive and configurable components and game elements. We showed that these adaptation elements allow for personalized interventions tailored to individuals with varying level of impairments, ranging from stroke patients, to older adults and to children with hemiplegia, overall aged 3 to 77 years old. (3) We investigated the effectiveness of our tangible robot-mediated activities with chronic stroke patients, and showed the effect of gamification on performance outcomes as well as patients' preferences via a controlled study. We showed that gamification leads to a more controlled motor performance and discovered that there are patients who do not like gamification, or for whom gamification is not well suited. (4) We extended the application of our approach into the context of healthy aging, investigating the effect of the Tangible Pacman Game and its game elements on the motor performance of the elderly. We explored age-related differences within a controlled study with a young control group, and showed positive motor learning and skill transfer. (5) We developed further variants of our proposed system with additional adaptations and features to increase its effectiveness and acceptance by: (5a) Designing dynamic game spaces with rearrangeable tiles, (5b) Incorporating IMUs to our games to detect and prevent compensatory motions, (5c) Developing co-located and remote multiplayer game modalities to promote social interaction, and elevating the tangible robots to the online space to bridge the patient's interaction with therapists or family members in the context of recent social distancing rules. Throughout this thesis, we aimed to strike a balance between exploration and exploitation. In the exploration side, we interleaved design, development and testing in multiple steps involving 7 therapy centers in 5 cities in 2 countries. This allowed us to successfully introduce and adapt our system to different therapy settings and persons affected by a large range of neurological difficulties and impairment levels. In the exploitation side, we conducted controlled studies, showing the effectiveness of our interventions on real target users. In total, the designs in this thesis touched 122 healthy participants, 96 target users and 18 domain professionals.