This study was to investigate the role of students' game self-efficacy in their game performance as well as in their visual behavior distributions and transfer patterns during their gameplay. A total of 48 university students were recruited for participating in a gameplaying task while wearing with an ASL Mobile Eye eye-tracker. After the gameplay, each participant was asked to report their self-efficacy of the gameplay. Based on the self-reported game self-efficacy (GSE) score, the participants were divided into the higher and lower GSE groups. Independent t tests were used to compare their game performances and eye-tracking indices for each area of interest. A lag sequential analysis was used to profile the visual transfer pattern for each group. The results showed that the higher GSE group had better performances than the lower GSE group. During the gameplay, the higher GSE students tended to pay more attention to the critical information for a successful play. In addition, the lag sequential analyses revealed different visual transfer patterns for the two GSE groups. The higher GSE group tended to show a critical-factor-centralized visual-trasfer pattern which demonstrated more planning and monitoring gameplaying strategies; while the lower GSE students, probably lacking the self-confidence of gameplaying, tended to pay more attention to the remaining time for the game.This study may provide suggestions for future game-based leraning system designs.