The neural mechanisms of hierarchical stimulus processing were investigated using a combined event-related potentials (ERPs) and positron emission tomography (PET) approach. Healthy subjects were tested under two conditions that involved selective or divided attention between local and global levels of hierarchical letter stimuli in order to determine whether and where hemispheric differences might exist in the processing of local versus global information. When attention was divided between global and local levels, the N2 component of the ERPs (260- to 360-msec latency) elicited by the target stimuli showed asymmetries in amplitude over the two hemispheres. The N2 to local targets was larger over the left hemisphere, but the N2 to global targets tended to be slightly larger over the right hemisphere. However, the shorter-latency, sensory-evoked P1 component (90- to 150-msec latency) was not different for global versus local targets under conditions of divided attention. In contrast, during selective attention to either global or local targets, asymmetries in the N2 component were not observed. But under selective attention conditions, the sensory-evoked P1 components in the extrastriate cortex were enlarged for global versus local attention. In- creased regional cerebral blood flow in the posterior fusiform gyrus bilaterally was observed in the PET data during selective attention to either global or local targets, but neither these nor the P1 component showed any tendency toward hemispheric difference for global versus local attention. Neither were there any activations observed in the parietal lobe during selective attention to global versus local targets. Together these data indicate that early sensory inputs are not modulated to gate global versus local information differentially into the two hemi- spheres. Rather, later stages of processing that may be asym- metrically organized in the left and right hemispheres operate in parallel to process global and local aspects of complex stimuli (i.e., the N2 effect of the ERPs). This pattern of results supports models proposing that spatial frequency analysis is only asymmetric at higher stages of perceptual processing and not at the earliest stages of visual cortical analysis.