The products of unisensory and multisensory integration within the Superior Colliculus were found to be appreciably different. Moreover, recent data from cats suggests that multiple stimuli from the same sensory modality only marginally enhance localization compared to cross-modal stimulus combinations. In the present study, we investigated whether the integration of stimuli from different modalities (cross-modal) and from the same modality (within-modal) have a different impact on spatial orienting in humans. To this aim, we asked subjects to perform a simple reaction time task (Experiment 1) and a localization task (Experiment 2), which comprised modality-specific stimuli (visual or auditory), cross-modal stimulus pairs (visual-auditory) and within-modal stimulus pairs (visual-visual). Although both the integrative modes shortened RTs compared to the best unimodal condition, the redundancy gain was significantly greater for cross-modal than within-modal stimulus combinations. Moreover, a violation of race model inequality was observed only for the cross-modal condition. In addition, cross-modal stimulus combinations yielded a greater improvement in stimulus localization, according to a Bayesian model of spatial integration. The present results suggest that the integration of stimuli from different modalities and from the same modality have a different impact on covert and overt orienting, and support the hypothesis that the behavioural products derived from multisensory integration are not attributable to simple target redundancy.