Bistable perception to ambiguous figure is an intriguing visual phenomenon. The underlying mechanisms, however, remain largely unclear. We addressed the issue by recording eye saccade and pupillary reflex in the perceptual responses of human subjects to the ambiguous figure and disambiguous figure (control) generated from the structure-from-motion stimuli. Their pupils dilated when the subjects reported the perceptual reversal between the two mutually exclusive states to both the ambiguous figure and disambiguous figure. The pupillary dilation reached the peak after the perceptual reversal. In contrast to the disambiguous figure, before reported the reversal to the ambiguous figure, the pupils were smaller than the mean size, while after the peak of pupillary dilation occurred, the pupils were still larger than the mean size. These results illustrate that the pupillary dilation posterior to the perceptual report can be regarded as an indication of the perceptual reversal that had occurred, while the differences in pupillary reflex to the ambiguous figure and disambiguous figure prior to the perceptual report probably reflects the intrinsic neural trace of the perceptual reversal and of the perceptual states. Additionally, in the process of perceiving structure-from-motion stimuli, the distribution in directions of eye saccades of subjects changed with the change in the motion axis of the stimuli. The change fashions were accordant between the ambiguous figure and disambiguous figure. This suggests that subjects have the same perception to the ambiguous figure and disambiuous figure and that the correlation of the pupillary reflex changes with the perceptual reversals to the bistable figure of structure-from-motion is reliable. The study casts new light on the visual mechanisms for bistable perception to ambiguous figure.