This study aimed to juxtapose the impacts of the Problem-Based Learning (PBL) methodology implemented within a three-dimensional (3D) virtual environment against PBL in a conventional face-to-face setting, along with a control group, on students' learning performance, conceptual comprehension, and spatial aptitude. The investigation concentrated on the structure of matter within a seventh-grade science curriculum, encompassing a cohort of 79 students enrolled in the course. The study was conducted under a mixed methods experimental design and comprised three distinct groups: two experimental groups (E1; E2) and one control group (C). The instructional intervention in the E1 group involved utilizing a 3D virtual environment grounded in the problem-based learning method, whereas the E2 group underwent face-to-face instruction employing worksheets derived from problem-based learning. In contrast, the control group received traditional direct instruction from the science teacher through lectures, demonstrations, and question-and-answer sessions, all focusing on the same topic. Pretests and posttests were administered to the participants before and after the experimental interventions, assessing conceptual understanding, spatial visualization, and mental rotation. Additionally, learning tasks completed by the E1 and E2 groups were evaluated using rubrics to assess learning performance. Subsequent to the post-test, individual semi-structured interviews were conducted with randomly selected students from the E1 group. The results indicated that instruction within the problem-based 3D virtual environment significantly enhanced students' learning performance, conceptual understanding, spatial visualization, and mental rotation skills compared to the other instructional conditions.