There are many tasks in which people are called on to disregard information that they have already processed. Dealing with inadmissible evidence in a courtroom setting, second-guessing the past, and responding to experimental psychologists’ debriefing instructions are 3 tasks of this type; in all these cases, people have been found to experience considerable difficulty. The present experiments investigated these difficulties in a general form, using almanac-type questions; 202 Ss participated. Those who were told the correct answers to such questions overestimated both how much they would have known about the answer had they not been told and how much they actually did know about the answer before being told. Attempts to undo this “knew-it-all-along” effect by exhorting Ss to work harder or telling them about the bias failed. Results are discussed in terms of how the structure of one’s knowledge is altered to accommodate new information.