Adaptativity is a key-concern when developing serious games for learning purposes. It makes it possible to customize the game according to each learner individuality. To deal with adaptativity, this chapter proposes a Model-Driven Engineering approach that supports dynamic scenarization instead of implementing fixed configurations of learning scenarios. The base principle is to consider the generation of scenarios as a model transformation of a learner profile and a game description models toward adapted scenarios. This proposal has been applied to the context of the Escape-it! research project that aims to propose an “escape-room” game for helping children with Autistic Syndrome Disorder (ASD) to learn visual performance skills.