Benzodiazepine agonists have been shown to reduce cognitive functioning by producing attentional de®cits similar to those produced by muscarinic antagonists. It has therefore been postulated that benzodiazepine antagonists may improve cognitive function in patients with Alzheimer's disease by indirectly increasing cholinergic function. Eleven patients with mild to moderate Alzheimer's disease took part in a double-blind placebo-controlled crossover study of the eects of a single 1 mg injection of the benzodiazepine antagonist ¯umazenil on cognitive function. Validated computer tests were used to assess change in cognitive function over time after each injection. No adverse eects of the drug were experienced. Speed on both a simple reaction time task and a picture recognition task was signi®cantly slowed by ¯umazenil compared to placebo at 15 min ( p 0 . 027 and p 0 . 01). Accuracy of recalling pictures was signi®cantly reduced compared to baseline ( p 0 . 0002), but this was not statistically signi®cant when compared to placebo. Possible reasons for these ®ndings are discussed and it is concluded that a single injection of 1 mg ¯umazenil produces cognitive slowing in patients with Alzheimer's disease. It is suggested that further research is needed into central benzodiazepine receptor density and function, as well as dose±response studies using ¯umazenil.