Extant research reports differential effects related to the efficacy of video games as a means to enhance science instruction. However, there are very few studies examining differences in learning outcomes across student-level independent variables. This study used multilevel modeling to examine the effects of three video game-enhanced life science units on the performance of 366 adolescent students from three middle schools in the central and western United States. The study included 18 life science classrooms that were taught by four teachers over a 6-week period. Independent variables included the students' National Assessment of Educational Progress reading level, presence of a learning disability, gender, and student perceptions of science and video games. Struggling readers and students with low perceptions of their skills as scientists underperformed when compared to their peers. Results across all games indicated significant learning gains for students with and without learning disabilities. Learning disability, gender, and perceptions of video games for learning were not significant predictors of student performance. Results from the study indicate that the effects of video games on science learning are mediated by individual learner characteristics and the science content.