Temporal binding is considered as an illusion of timing perception, in which an action (e.g. a keypress) and its action‑effect (e.g. a sound with a 250 ms delay after the keypress) are perceived to temporally attract towards each other. The event timing perception is usually measured by asking participants to report the position of a fast‑rotating clock hand at the time of event onset. For example, when reporting the time of a sound onset, the reported clock hand was at an earlier position when it was preceded by a keypress than when not, which was taken as the evidence that the sound was perceived to occur earlier when it was triggered by a keypress than when presented alone. We show here that the distribution of visuospatial attention could explain the temporal binding effect. In the case of making a keypress to trigger a sound, stronger visual attention was found at early clock hand positions when judging the sound time and at late clock hand positions when judging the action time, compared to when the keypress or the sound was presented alone. Furthermore, experimentally manipulating the distribution of visual spatial attention resulted in predicted changes in timing reports. Finally, the temporal binding effect could be reproduced through cognitive modelling using the attention data alone as input. Our results suggest that visuospatial attention is at least a confound that should be controlled for before any conclusion about timing perception can be drawn when using the clock method.