In the present study, we investigated whether a hemispheric division of labor is most advantageous to performance when lateralized inputs place unequal resource demands on the left and right cerebral hemispheres. In each trial, participants decided whether 2 rotated letters, presented either in the same visual field (within-field trials) or in opposite visual fields (across-field trials), were both of normal orientation, or whether one was normal and the other was mirror-reversed. To discriminate a letter's orientation, one must rotate the letter to the upright position. Therefore, we manipulated whether the two letters imposed similar or dissimilar demands on cognitive resources by varying the number of degrees that each letter needed to be rotated to reach the upright position. As predicted, in 2 experiments we found that the across-field advantage increased as the number of degrees each letter needed to be rotated became more dissimilar. These findings support a current model of hemispheric interactions, which posits that an unequal hemispheric distribution of cognitive load allows the cerebral hemispheres to take the lead for different aspects of cognitive processing.Although it is widely recognized that the left and right cerebral hemispheres process information quite differently (Hellige, 1993;Springer & Deutsch, 1998), relatively little is known about how interactions between the cerebral hemispheres modulate cognitive processing. Recently, Banich and colleagues (Banich, 1998;Banich & Belger, 1990) have proposed that a division of processing across the hemispheres increases the processing capacity of the brain. According to Banich (1998), a division of critical information across the hemispheres allows for a more efficient hemispheric division of labor than does directing critical information to a single hemisphere. In support of this view, Banich and colleagues have consistently demonstrated that as the resource demands for a task increase, across-field processing (i.e., a division of critical information across the left and right visual fields) becomes more advantageous to performance relative to withinfield processing (e.g