Motor rehabilitation is known to have several difficulties regarding patient's engagement and therapy adherence. Since advances in physiological computing technologies have had an exponential grown in the last decade, its use in novel therapies for motor rehabilitation is being popularized. Serious games, for instance, have been applied as a complementary therapy for neuromuscular disorders, being the game design process a key factor to influence both the attractiveness and effectiveness of the game. In this paper, we expose a design methodology used for the creation of a serious videogame for motor rehabilitation of upper limbs using surface electromyography (sEMG) as the human-computer interface to control the game and monitor the players' fatigue levels. By using an adaptation mechanism from the physiological computing field, called biocybernetic adaptation; the videogame can adapt the game difficulty based on measured fatigue levels. The game design was also informed with therapeutic recommendations and followed an iterative design process. We hope this paper can reveal important insights for both engineers and game designers to create more physiologically intelligent solutions for motor rehabilitation.