The aim of this study was to investigate which brain regions are involved in the integration of multisensory object information. Familiar objects were presented in an S1-S2 paradigm (in which two stimuli - S1 and S2 - are presented consecutively with a short time delay). The modality of S1 was varied blockwise (visual, haptic, auditory), while S2 was always auditory. Within a trial, S1 and S2 either referred to the same object or to two different objects. Thirty-one participants took part in this event-related fMRI study. We compared BOLD responses in object-incongruent and object-congruent trials. In the two crossmodal conditions the dorsolateral prefrontal cortex, the insula and the anterior cingulate cortex (ACC) were significantly more activated in object-incongruent than in object-congruent trials. Parts of the ACC were activated in the unimodal condition as well, suggesting that the ACC activation was due to incongruence of S1 and S2 independent of the modality of the objects. Additionally, we found specific activation for each crossmodal condition: in visual-auditive trials, the superior colliculi and in haptic-auditive trials, the intraparietal sulcus were more activated in object-incongruent than in object-congruent trials. Our data provides evidence for the interaction of multiple brain sites in multisensory object recognition.