In product design education, identifying proportion and manipulating proportional relationships are important practices in form-giving training. However, using conventional 2D displays with monocular depth cues to train students with different spatial abilities remains a great challenge. Although, some literature has indicated that stereoscopic 3D (S3D) displays were helpful for depth-related tasks, whether S3D is helpful for proportion estimation is an open question. Therefore, the objective of this research was to study whether using S3D for design department students could assist them in improving their ability to interpret the proportions of products. The independent variables of the experiment included spatial ability, task complexity, and display mode. Spatial ability was a between-subject variable. Students aged 20–25 years were recruited as participants and were categorized into high-, moderate-, and low-performance groups depending on their score on a spatial ability scale. Task complexity and display mode were within-subject variables. In the experiment, three chair styles were used as samples. Digital models of three chairs with distinct proportions were displayed in two conditions: 2D mode with monocular depth cues and S3D mode. Participants were asked to compare the proportions of a physical chair and three digital chairs and then select the digital chair with the correct proportions in the 2D and S3D modes. The dependent variable were the score of selecting the 3D digital models with correct proportions compared to the physical objects. The result indicated that students could perceive proportions more accurately in the S3D mode than in the 2D mode. In particular, when using S3D, participants with low spatial ability demonstrated overall performance that was equal to that of participants with high spatial ability.