We designed a memory task involving visual recognition sensitive enough to demonstrate the long-term amnesic effect of diazepam. Following memorization of a list of abstract visual stimuli (Fourier), subjects were evaluated on recognition performance immediately following acquisition and after a 3-day drug clearance interval. Administration of 15 mg diazepam 1 h before acquisition imposed a significant (10-24%) deficit only in delayed recognition. In contrast, a drug-free acquisition followed by a diazepam-challenged delayed recognition did not influence recognition. Moreover, 1 h after administration, diazepam did not significantly impair detection or visual discriminative performances. Given the persistence of an important deficit in recognition memory in the absence of any drug, this paradigm promises to be useful for studying regional cerebral blood flow during long-term memory performance.