Benzodiazepines (BZs) affect acquisition of new information, while retrieval of already learned information is unimpaired. The variability of this effect is important and depends on the nature of the BZ, its dose, the route of administration and the susceptibility of the subject taking the drug. This last factor depends itself on the anxiety level, the age, and a less known idiosyncratic susceptibility of the patient. Finally, there is probably a partial tolerance for the amnestic action of BZs, which explains the fact that the most dramatic amnesias have been described after administration of a single dose of BZ, taken by a patient unaccustomed to BZs. The value of pharmacocinetic and pharmacodynamic characteristics in predicting cognitive impairment remains misunderstood, even though in clinical practice the greatest amnestic effects have been described with short-acting BZs. The interest of studying BZs induced amnesia rely upon several arguments: first, it can be an harmful side-effect, which could be avoided or at least predicted by a better knowledge of BZs and the synthesis of new and more specific drugs; secondly it is an interesting model of organic amnesia, which could allow a better understanding of normal memory.