A variety of computer-based 3D applications are becoming regular tools for dental students for self-learning. This study investigated the learning effectiveness of junior dental students in passively versus actively controlling the 3D virtual scenes of implant dentistry. Participants were randomized into three groups and were exposed to three designs of educational materials: traditional 2D webpages (2D); active-controlling 3D webpages (A3); and passive-controlling 3D webpages (P3). After reviewing the webpages, the participants were asked to complete a posttest to assess the relative quality of information acquisition. Their responses were compared and analyzed. The results indicated that the P3 group received the highest score of 26.4+/-3.1 on the posttest, significantly better than the A3 group, which had the worst performance with a score of 20.3+/-4.0. The 2D group received a score of 24.2+/-4.6. There was a significant correlation between the scores on a mental rotations test and the subjects' performance on the posttest (p<0.001). A serious disadvantage of active control was indicated for individuals with low spatial ability. In 3D virtual reality assisted self-learning, passive control produces higher learning effects compared to active control. Too much active control may generate significantly negative impacts on students, especially for individuals with low spatial ability.