The science of serious games is growing at a substantial pace, providing new insights into the nature of game-based learning. Recently, research has begun to focus on the elements that comprise serious games and how these elements relate to learning (Wilson et al., 2009; Pavlas et al., 2009). As part of an effort to understand how these attributes impact learning outcomes, a study manipulating a number of game attributes in an immune system game was conducted. From this effort, two psychological constructs initially considered as mere covariates – video game self-efficacy and flow state – emerged as significant and highly explanatory predictors of learning. This article provides an overview of these constructs, describes the research that led to this finding, presents the results of this research, and offers implications and suggestions for future work.