Unlike most research, which has primarily examined the players' interest in or attitude toward game-based learning through questionnaires, the purpose of this empirical study is to explore students' engagement patterns by qualitative observation and sequential analysis to visualize and better understand their game-based learning process. We studied the sequential behaviors of 34 students (17 male and 17 female) and considered issues of gender differences and sequential pattern similarities. The results show that the behavioral coding schema provided by the authors and the innovative method of sequential analysis can provide researchers with a certain level of understanding of students' engagement patterns in game-based learning environments. In terms of the overall sequence results, this study identified higher and lower engagement patterns to represent students' learning processes in game-based learning. Moreover, the sequential patterns represent qualitative differences and similarities in engagement patterns grouped by gender. A good engagement cycle, in which male and female students started the game to attempt to think and solve problems, was noticeable. However, male students were observed to demonstrate more engaged behaviors, with continuous self-conversations when confused. The frequency of self-conversation from female students was obviously lower than that of male students and revealed more verbal and nonverbal behaviors. The deep examination of students' verbal and nonverbal engagement behaviors may make beneficial contributions to the educational technology field with the adoption of sequential analysis.