Abstract : When we feel certain about our factual knowledge, all too often we are wrong. This phenomenon, labeled 'the certainty illusion', is demonstrated in four experiments in which subjects (1) answered questions about a variety of topics and (2) indicated their degree of certainty about each answer. Subjects were wrong frequently on answers judged certain to be correct. Careful tutoring of subjects in the subtleties of expressing their certainty in terms of probabilities and odds did little to reduce the illusion. Feelings of certainty were so strong that subjects were willing to bet on the correctness of their knowledge. Because of the illusion, the bets they accepted were quite disadvantageous to them. The psychological basis for unwarranted certainty is discussed in terms of the inferential processes whereby knowledge is reconstructed from fragments of perceptions and memories.