The purpose of this study was to explore how students interacted with guidance to conduct a scientific inquiry in a physics simulation by using the eye-tracking techniques. The participants were 51 7th graders, and an eye-tracking system was used to record their visual behaviors and log data while they were using the simulation. As for data analysis, we first checked each participant's log data to examine whether they completed the requirement of the guidance, and then checked the correctness of her/his answer to the inquiry task. The participants were thus divided into two groups (correct vs. wrong), and the patterns of their visual behaviors were examined by a set of eye-movement indices, normalized heat maps and lag sequential analyses. The results indicate that both spatial distributions and temporal sequences of the participants' visual attention were associated with their performances on the inquiry task. Regarding the spatial distribution, the correct group tended to allocate more visual attention to the regions presenting the target phenomenon than the wrong group. Concerning the temporal sequence, the correct group tended to make more visual transitions among the content of the guidance, the relevant control panels and the target phenomena than the wrong group.