The neural correlates of gist and verbatim encoding of episodes sharing semantically related components were investigated. At study, participants were presented with lists of word pairs formed by pairing one of two associated words with different associates of an unstudied theme word. At test, participants discriminated Old pairs from rearranged pairs generated by re-pairing study pairs belonging to the same study list (Intra pairs) or different study lists (Inter pairs). ERPs were recorded for each study pair. For the study pairs that later served as Old pairs, the ERPs were more positive when these pairs were subsequently correctly identified than when they were incorrectly rejected. For the study pairs containing words that subsequently appeared in Intra pairs, the ERPs were also more positive when the Intra pair was correctly rejected than when it was incorrectly accepted. However, the topographies of the subsequent memory effects for accepted Old pairs and rejected Intra pairs were different. We argue that specific information is more important for rejecting Intra pairs than accepting Old pairs. The different topographies of these two subsequent memory effects therefore reflect qualitatively distinct encoding operations for specific and nonspecific information. (C) 2010 Elsevier B.V. All rights reserved.