We reanalyzed the behavioral and fMRI data from seven previously published studies of working memory in order to assess the behavioral and neural effects of item-nonspecific proactive interference (PI; attributable to the accrual of antecedent information independent of the repetition of particular items). We hypothesized that item-nonspecific PI, implicated in age-related declines in working memory performance, is mediated by the same mechanism(s) that mediate itemspecific PI (occurring when an invalid memory probe matches a memorandum from the previous trial). Reaction time increased across trials as a function of position within the block, a trend that reversed across the duration of each multiblock experiment. The fMRI analyses revealed sensitivity to item-nonspecific PI during the probe epoch in the left anterior inferior frontal gyrus and the left dorsolateral prefrontal cortex (PFC). They also revealed a negative trend, across trials, in the transient probe-evoked component of the global signal. A common PFC-based mechanism may mediate many forms of PI.Working memory refers to the cognitive capacity that enables the temporary on-line maintenance and manipulation of information when that information is no longer present in the environment. Because working memory has been implicated as a critical contributor to such complex cognitive abilities as language comprehension, learning, planning, reasoning, and general fluid intelligence (Baddeley, 1992;Engle, Kane, & Tuholski, 1999;Jonides, 1995;G. A. Miller, Galanter, & Pribram, 1960), understanding the factors that govern its success or failure under various conditions is an important goal of cognitive neuroscience. Although the causes of forgetting in working memory are not fully understood, decades of psychological research indicate that interference (more so than, for example, decay associated with the mere passage of time) is an important factor. Thus, Keppel and Underwood (1962) famously demonstrated that there is virtually no forgetting on the first trial of the Brown-Peterson task, regardless of the length of time separating target presentation from response, but that errors are already present on the second trial, even at the shortest of delays. And Wickens and colleagues have established that proactive interference (PI)-the disruption of performance on a memory task attributable to antecedent information -need not be produced by the repeated presentation of one or more particular items but that it is also present if items from preceding trials are drawn from the same class of stimuli (e.g., letters or digits; Wickens, Born, & Allen, 1963) or from the same semantic category (in the case of nouns), or even if they are simply presented with similar perceptual characteristics (e.g., letter case, modality, figure/ground, or slide area; Wickens, 1973). The present study Correspondence concerning this article should be addressed to B. R. Postle, 1202 W. Johnson Street, Madison, WI 53706 (e-mail: postle@wisc.edu).. This research was supported by NIH G...