The objective of this study is to investigate whether strategy differences in mental rotation performance can be explained using gaze durations collected from an eye tracker. To investigate this, 36 participants were administered an electronic version of the Purdue Spatial Visual Test: Rotation. Three questions (out of 30) were selected based on the overall response accuracy to represent an easy, moderately difficult, and difficult question. Participants were then segmented into two groups (high accuracy and low accuracy) based on test scores. Significant interaction effects between the groups and the five possible choices were observed for the gaze durations of each question (easy: F(4,170) = 9.05, p < 0.0001; moderately difficult: F(4,170) = 41.12, p < 0.0001; difficult: F(4,170) = 2.89, p< 0.05). The gaze durations of the high accuracy group were primarily skewed towards one of the five potential response choices suggesting a holistic approach. For the low accuracy group, an analytic approach appears likely with gaze durations more evenly distributed across all choices.