This work sought to help inform the design of educational digital games by the studying the design of successful commercial videogames. The main thesis question was: How does a commercially and critically successful modern video game support the learning that players must accomplish in order to succeed in the game (i.e. get to the end or win)? This work takes a two-pronged approach to supporting the main argument, which is that the reason we can learn about designing educational games by studying commercial games is that people already learn from games and the best ones are already quite effective at teaching players what they need to learn in order to succeed in the game. The first part of the research establishes a foundation for the argument, namely that accepted pedagogy can be found in existing commercial games. The second part of the work proposes new methods for analysing games that can uncover mechanisms used to support learning in games which can be emplyed even if those games were not originally designed as educational objects. In order to support the claim that ‘good’ commercial videogames already embody elements of sound pedagogy an explicit connection is made between game design and formally accepted theory and models in teaching and learning. During this phase of the work a significant concern was raised regarding the classification of games as ‘good’, so a new methodology using Borda Counts was devised and tested that combines various disjoint subjective reviews and rankings from disparate sources in non-trivial manner that accounts for relative standings. Complementary to that was a meta-analysis of the criteria used to select games chosen as subjects of study as reported by researchers. Then, several games were chosen using this new ranking method and analysed using another new methodology that was designed for this work, called Instructional Ethology. This is a new methodology for game design deconstruction and analysis that would allows the extraction of information about mechanisms used to support learning. This methodology combines behavioural and structural analysis to examine how commercial games support learning by examining the game itself from the perspective of what the game does. Further, this methodology can be applied to the analysis of any software system and offers a new approach to studying any interactive software. The results of the present study offered new insights into how several highly successful commercial games support players while they learn what they must learn in order to succeed in those games. A new design model was proposed, known as the 'Magic Bullet' that allows designers to visualize the relative proportions of potential learning in a game to assess the potential of a design.