This study examined the development and implementation of a transfer activity for a classroom game-based learning environment called CRYSTAL ISLAND - UNCHARTED DISCOVERY. The game focused on science curriculum including map models, map navigation, and landforms for 5th grade students. Students interacted with CRYSTAL ISLAND - UNCHARTED DISCOVERY during twelve 50-minute sessions in which six class periods involved playing the game and six involved supplemental lessons that aligned with the curriculum in the game. Students (N = 594) across eight schools participated. The transfer activity required students to create an island on grid paper that included at least seven different landforms, a map navigation activity created for peers, and an attached map scale model. The activity included all three curricular objectives found in CRYSTAL ISLAND - UNCHARTED DISCOVERY. Findings revealed that students showed a significant overall increase in content knowledge after the gameplay sessions. Prior knowledge in science and change in pre- to posttest knowledge predicted scores on the transfer task. Interestingly, video game self-efficacy was a significant negative predictor of transfer performance. When other variables were accounted for, none of the measured motivational variables including interest for the game, performance-approach goal orientation, mastery-approach goal orientation, or self-efficacy for science were significant predictors.