In two experiments, participants viewed a videotape of a simulated armed robbery, later answered misleading questions about it, and then finally completed a source monitoring test. For the test, participants were asked to indicate for each test item whether it was (1) seen in the video only, (2) read about in the questions only, (3) both seen and read about, (4) not remembered or (5) known to have occurred but the source was unclear. The latter response category was included on the test to remove source guessing and to ensure that attributions to ‘video’, ‘questions' or ‘both’ were caused by false conscious recollection. In Expt 1, robust misinformation effects were obtained with both 1- and 48-hour delays between receiving misinformation and the memory test. However, suggested objects were more likely to receive ‘video only’ attributions at long delay than at short. Experiment 2 verified that it was the interval between receiving the misinformation and the test, and not the interval between viewing the video and receiving the misinformation, that determined the effect of delay in Expt 1. The results are explained by assuming that, at short delay, participants remembered reading about the suggested objects and could discount the ‘video only’ category. However, despite accurately remembering the source of suggested information, the misinformation effect as measured by ‘both’ responses was not diminished. Thus, remembering that misinformation was suggested does not necessarily stop the creation of false memories.