The use of videos for review and preview has been widely adopted as a strategy to complement classroom instruction. The purpose of this study is to analyze elementary school students’ attention and gaze allocation in science review and preview situations using video-based learning by extracting mind wandering (MW) and gaze attention through eyetracking methods. The study participants were 13 sixth-grade elementary school students. The tasks for review and preview were selected based on the class curriculum and involved editing EBS educational videos into 10-minute segments. Data collection was conducted using Tobii’s X2-60 eye-tracking device. The review task was performed after the classroom lesson, whereas the preview task was performed before the classroom lesson. For data analysis, MW occurrences were identified and extracted based on fixation durations of over 400 ms that lasted for at least 2 seconds. Fixation duration for AOIs (Areas of Interest) was also extracted and analyzed. The results showed that MW, indicating disengagement of attention, occurred at an average rate of 3.09% in the review task and 2.54% in the preview task, with no significant difference in MW time and frequency between the two conditions. Additionally, there was no significant difference in MW time between the review-preference and preview-preference groups during task performance. The frequency of MW occurring in each TOI (Time of Interest) segment of the review and preview tasks also showed no significant difference. Regarding gaze attention allocation, students paid more attention to lecture slides in the review condition, while they focused more on the teacher in the preview condition. These findings suggest that although there is no significant difference in the level of attention between review and preview tasks, different cognitive processing patterns emerge in each condition.