In virtual prosthetic training research, serious games have been investigated for over 30 years. However, few game design elements are used and assessed for their effect on the voluntary adherence and repetition of the performed task. We compared two game-based versions of an established myoelectric-controlled virtual prosthetic training task with an interface without game elements of the same task [for video, see (Garske, 2022)]. Twelve limb-intact participants were sorted into three groups of comparable ability and asked to perform the task as long as they were motivated. Following the task, they completed a questionnaire regarding their motivation and engagement in the task. The investigation established that participants in the game-based groups performed the task significantly longer when more game design elements were implemented in the task (medians of 6 vs. 9.5 vs. 14 blocks for groups with increasing number of different game design elements). The participants in the game-based versions were also more likely to end the task out of fatigue than for reasons of boredom or frustration, which was verified by a fatigue analysis of the myoelectric signal. We demonstrated that the utilization of game design methodically in virtual myoelectric training tasks can support adherence and duration of a virtual training, in the short-term. Whether such short-term enhanced engagement would lead to long-term adherence remains an open question.