The “ventriloquist effect” refers to the fact that vision usually dominates hearing in spatial localization, and this has been shown to be consistent with optimal integration of visual and auditory signals (Alais and Burr in Curr Biol 14(3):257–262, 2004). For temporal localization, however, auditory stimuli often “capture” visual stimuli, in what has become known as “temporal ventriloquism”. We examined this quantitatively using a bisection task, confirming that sound does tend to dominate the perceived timing of audio-visual stimuli. The dominance was predicted qualitatively by considering the better temporal localization of audition, but the quantitative fit was less than perfect, with more weight being given to audition than predicted from thresholds. As predicted by optimal cue combination, the temporal localization of audio-visual stimuli was better than for either sense alone.