Predicted motion (PM) tasks test the accuracy of predicting the future position of a moving target. Previous PM studies using audiovisual stimuli have suggested that observers rely primarily on visual motion cues. To clarify the role of auditory signals in predicting future positions of bimodal targets, we designed a novel PM task where spatial coincidence of audio and visual motion signals was varied in three conditions: auditory and visual motion stimuli were spatially correlated (congruent condition), the auditory motion stimulus was moving behind the visual motion stimulus (sound-trailing condition), or the auditory motion stimulus was moving ahead the visual motion stimulus (sound-leading condition). We manipulated target speed (5.5 or 11 cm/s), the time that the moving audiovisual stimulus was presented (500 or 750 ms viewing time), and the time the visual stimulus disappeared while the auditory stimulus continued to move by itself before prompting subjects to estimate the position of the visual stimulus would have traveled if it continued along with the auditory stimulus (750, 1,000, or 1,500 ms prediction time). We also included two unimodal control conditions: visual-only and auditory-only. Subjects (n = 12) typically overestimated the target position of congruent bimodal targets. In the sound-trailing and sound-leading conditions, pointing responses were biased in the direction of the auditory stimulus, showing that PM performance is not reliant solely upon visual motion cues. We conclude that putative cognitive extrapolation mechanisms assume spatial coherence of bimodal motion signals and may perform some averaging of these motion signals when they do not spatially coincide.