Information from the auditory and visual systems converges in the nervous system with physiological and behavioral consequences. Most of our knowledge about the rules governing such convergence has been obtained in experiments where the strength or the timing of the individual auditory and visual stimuli has been varied. Relatively little attention has been paid to the spatial relationship between different modalities of stimuli in multisensory experiments. We studied saccadic reaction times of human subjects to bimodal auditory and visual stimulus presentations under two conditions: first, with the targets spatially coincident and, second, with various degrees of spatial separation or disparity. In the first experiment, we found that the saccadic reaction times were consistently shorter than would be predicted by independent processing of information about the visual and auditory targets. These results suggest convergence of multimodal information at one or more loci within the nervous system. In the second experiment, we found that saccadic latency gradually increased as spatial distance between the auditory and visual targets increased. Evidence for neural summation was found over a wide range of spatial disparities. These results suggest that multisensory information can be integrated and have significant influences on behavior over a surprisingly large range of spatial disparity.