Selective attention has been hypothesised to reduce distractor interference at both perceptual and post-perceptual levels (Lavie, 2005), respectively by focusing perceptual resources on the attended location and by blocking at post-perceptual levels distractors that survive perceptual selection. This study measured the impact of load on these two levels of selection using a flanker paradigm (Eriksen & St. James, 1986) and indexing distractor interference as a function of separation. It distinguished changes in the extent of focus of the distractorinterference function of separation (reflecting the perceptual level of selection) from changes in the amplitude of distractor interference not accompanied by changes in focus (reflecting the post-perceptual level of selection). It showed that: (1) the spatial profile of perceptual resources is shaped like a 'Mexican hat' (Müller et al., 2005); (2) increasing perceptual load causes perceptual resources to focus (Caparos & Linnell, 2009); (3) increasing cognitive load causes perceptual resources to defocus; and (4) participants with reduced working-memory span show reduced post-perceptual blocking of distractors. While these findings are consistent with two levels of selective attention, they show that the first perceptual level is affected not only by perceptual but also by cognitive-control mechanisms.