<p>This study explores the effectiveness of Virtual Reality technology in natural science courses (pH, Aqueous Solutions, and related topics) in Taiwan’s primary education. The instructors are using the K-12 Four-Element Learning Model by the Taiwan Ministry of Education (Self-learning, Intra-group learning, Inter-group learning, and Teacher-guided learning) to plan the development of VR teaching activities for the course.) This study was conducted with an experimental group (receiving VR-assisted instruction) and a control group (receiving multimedia instruction). The experimental group was randomly chosen from schools that expressed a willingness to participate in the study. A learning motivation questionnaire was developed based on the ARCS framework, assessing four dimensions: attention, relevance, confidence, and satisfaction. The findings reveal four key results. First, elementary school students showed a generally consistent learning attitude toward VR-integrated and traditional instruction in natural science courses, with no significant variation across grades (fourth, fifth, and sixth), suggesting that VR teaching offers stability and broad applicability. Second, the experimental group scored significantly higher than the control group across all four ARCS components, with the most noticeable differences in attention and satisfaction. This indicates that VR instruction is particularly effective in enhancing students’ engagement and their sense of connection to the course. Third, among students from the two groups across different grade levels (fourth, fifth, and sixth grades), only the Satisfaction dimension of the ARCS model reached statistical significance. This indicates an interaction effect between VR instruction and grade level on students’ satisfaction. Fourth, gender did not show significant differences in any of the ARCS components, meaning that it is not a determining factor in students’ VR learning performance. In conclusion, this study demonstrates that instruction with VR integration can effectively increase primary school students’ engagement in science learning and their sense of connection to the course. The approach holds practical value and educational significance. Teachers who participated in the experiment also affirmed the Four Learning Model and its potential for guiding the design of VR-based learning activities with meaningful educational applications.</p>