Confabulation—fabricated or distorted memories about oneself—occurs in many disorders, but there is no reliable technique for investigating it in the laboratory. The authors used hypnosis to model clinical confabulation by giving subjects a suggestion for either (a) amnesia for everything that had happened since they started university, (b) amnesia for university plus an instruction to fill in memory gaps, or (c) confusion about the temporal order of university events. They then indexed different types of memory on a confabulation battery. The amnesia suggestion produced the most confabulation, especially for personal semantic information. Notably, subjects confabulated by making temporal confusions. The authors discuss the theoretical implications of this first attempt to model clinical confabulation and the potential utility of such analogues.