Visual search involves perception, attention, memory, and response selection. In this EEG/ERP study, we investigate how these components change with the type of search. Participants identified the location of a target item (coloured, oriented rectangular bar) in search displays also containing non-target distractors. In the condition yielding efficient search, where response time was independent of set size (two or four items), distractors had no feature in common with the target. In the condition yielding inefficient search, where response time increased with set size, distractors shared one feature (either colour, or orientation) with the target. Stimulus- and response-locked event-related potentials (P1, N1, P2, N2, P3, and rLRP) were analyzed. P3, an indicator of working memory engagement, showed no significant difference in onset latency between efficient and inefficient conditions. Although P3 peak amplitude was smaller, activity over time was greater for inefficient search. The duration of response-locked lateralized readiness potential (rLRP), indicating response selection, was also longer in the inefficient condition. By contrast, indicators of selective attention, P1 peak amplitude and N1 peak latency, were sensitive to number of items; and P2 peak amplitude to number of non-target features. (N2 results were inconclusive.) These results suggest a difference between efficient and inefficient search that lies not in the time required to select items as potential targets of search (selective attention), but in the time required to confirm which of those items is indeed the target (target confirmation).