Serious games are a growing field in academic research and they are considered an effective tool for education. Game-based learning invokes motivation and engagement in students resulting in effective instructional outcomes. An essential aspect of a serious game is the method of support for presenting the teaching material and providing feedback. A support design that evaluates students' progress and adapts accordingly, has the potential of producing better learning results. This paper presents an adaptive model based on fuzzy logic that adjusts the support acquisition according to student knowledge level. A serious game for teaching the concepts of sequence and iteration in programming to novice students was built to assess the model. It employs working examples as a support method since previous research indicated that it produced less cognitive load during problem-solving. An empirical study with 102 students has been conducted to evaluate the learning efficiency of the model. The analysis indicates positive results and a potential solution for balancing the amount of assistance in serious games.