In order to behave effectively in a complicated world, people must be able to focus their attention on goalrelevant stimuli at the expense of irrelevant ones. However, it is also important that attention can be captured by irrelevant stimuli when they are unique and may, therefore, signal potentially important changes in the environment. A central line of attention research has investigated attentional capture by such unique yet task-irrelevant singleton stimuli.It has long been established that attentional allocation toward stimuli in nontarget locations produces performance costs, as has been shown in spatial-cuing studies (e.g., Jonides, 1980;Posner, Nissen, & Ogden, 1978), as well as in studies of attentional capture by a singleton item presented within a visual search display (e.g., Jonides & Yantis, 1988;Theeuwes, 1992). Both these areas of research address the consequences of paying attention to irrelevant spatial locations. More recently, research has begun to address the effects of attentional allocation to irrelevant temporal positions. However, very little previous research has investigated the possibility of attentional capture by stimuli appearing at irrelevant temporal positions. Here, we ask whether a unique yet irrelevant singleton can produce temporal attentional capture during a rapid serial visual presentation (RSVP) search task.A few recent studies have provided some evidence that attentional allocation to items presented at irrelevant temporal positions can lead to performance costs. For example, Folk, Leber, and Egeth (2002) have shown attentional capture in an RSVP task by color singleton distractors flanking the central RSVP letters. However, since the singleton distractors in this study were spatially separated from the targets, it is not clear whether the capture effects were due to diversion of attention to an irrelevant spatial location (i.e., spatial attentional capture) or to an irrelevant temporal position (i.e., temporal attentional capture), or both.Research into the attentional blink (AB) has shown that attending to (rather than ignoring) the first of two targets can prevent participants from detecting the second, as long as the second target occurs within 500 msec of the first (e.g., Broadbent & Broadbent, 1987;Raymond, Shapiro, & Arnell, 1992). Thus, attentional allocation toward one item in a serial stream can interfere with processing of an item at a different point in the stream. These results cannot be explained in terms of spatial attention, because all the items were presented in a central RSVP stream. However, since participants respond primarily to the first target and only later to the second target, the AB is likely to involve both response-and memory-related effects (e.g., Jolicoeur, 1998). Moreover, most of the AB research assesses the consequences of attending deliberately to target stimuli and, as such, does not provide information about the involuntary capture of attention.Recent findings from the AB paradigm that an ignored first target or additional singleton ...