Temporal binding refers to the perceived temporal compression between an action and its delayed sensory outcome. The Libet clock is the most widely used method for measuring the effect. In this method, participants watched a clock with a rapid rotating clock hand. The time of the event of interest was reported as the clock hand position when the event occurred. In the present study, we evaluated the role of eye movements in the time reports obtained with the Libet clock. We conducted a 2 (with eye movement control vs. without eye movement control) × 2 (action-sound vs. sound-only) within-subjects experiment. The task was to report the onset time of a sound. In the action-sound condition, the sound was triggered by a keypress of the participants with a 250 ms delay. In the sound‑only condition, the sound was controlled by the computer. The reported sound onset time was earlier in the action-sound condition than in the sound-only condition (p < 0.05), thus replicating the temporal binding effect. However, when the eye movements of participants were strictly controlled during the experiment (i.e. no eye movements were allowed), the temporal binding effect was smaller than when there was no eye movement control (p < 0.05). Further data analysis showed that the eye movement control influenced the time reporting in the action-sound condition, but not the sound-only condition. Our results suggest that eye movements play an important role in the measurement of temporal binding, possibly due to an influence of visuospatial attention.