The neural basis of spatial attentional control have mostly been examined within a unimodal setting while presenting separate blocks of exogenous and endogenous trials. Here, we investigated this issue in a more ecological manner by using a spatial double-cuing paradigm in a multisensory setting. Participants discriminated the elevation (up/down) of a visual target equiproblably presented to either side of fixation. Targets were preceded by a central-symbolic, spatially-predictive visual cue voluntarily driving their attention to one side, followed by a peripheral, spatially-nonpredictive auditory signal designed to reflexively capture their attention. We found that the typical cost in responding to invalid as compared to valid exogenous and endogenous signals was associated with the activation of superior parietal and dorsal fronto-parietal cortical areas, respectively. However, greater activations within these separate neural systems were found when both cues required an attentional shift (i.e., they were both invalid). Furthermore, a greater contralateral extrastriate response to the target was observed in valid trials when both cues coherently indicated the side of target occurrence. Altogether, these results highlight the existence of separated-but-interacting neural circuits for the multisensory control of spatial attention, showing how endogenous/exogenous multisensory signals jointly interact to control the locus of our attention in space.