Reaction time (RT) is usually speeded and slowed to correctly and incorrectly cued target stimuli, respectively, in experiments on attention shift (Posner, 1988). When targets occur in a spatial location different from the cue, shifting attention from the cued location to the target location involves interrupt of ongoing activity, move attention to the new location, and reengage attention. The present study investigated whether the cognitive operations involved in cued vs uncued attentional shifts also were detectable in the event-related potentials (ERPs). Two experiments are reported, one with visual and one with auditory stimuli. In the visual modality, the cue and target were a lit-up square and an asterisk, respectively, shown on a computer screen. In the auditory modality, the cue was a tone in either the left or right ear, followed by the target noise in either the same or the opposite ear. The results from the visual experiment showed shorter RTs to correctly cued (Valid) targets and longer RTs to incorrectly cued (Invalid) targets. Invalidly cued targets elicited enhanced P3 amplitudes, with a frontocentral distribution. Reduced P3 amplitudes were observed on Valid trials, largest reduction for parietooccipital leads. The auditory experiment showed P3 enhancement frontally and reduced amplitudes at temporal and parietal leads. No RT differences were seen to Valid and Invalid trials. The existence of an anterior attention system involved in interrupt and disengage of attention, and a posterior attention facilitation system related to cue presentation is discussed.