Serious digital for education are presumed to be engaging, but little is known about whether engagement is ubiquitous, whether it persists over time, whether it is found for all students across the full range of prior gaming experience, and whether it is actually associated with game-based learning outcomes. To address these gaps, student behavioral engagement (i.e., sustained attention, persistence was examined during mathematics instruction using a mathematics serious digital game for the iPad?, in a study of 97 11-14 year olds (i.e., 6-8th grade students in the United States system). Consistent with the study hypotheses, observations revealed that most students were highly engaged, but engagement was lower for students who were observed later in the semester and who had the most prior gaming experience. In addition, observed engagement was associated with better performance on an assessment of the skills being taught in the game. In contrast, none of these effects was evident for self-reported student engagement. Time course, prior gaming experience, and method of assessing engagement are important modulators of student differential response to game-based instruction and should be considered when evaluating the influence of serious digital games on learning outcomes.