The effects of single doses of chlorpromazine (100 mg) and lorazepam (0.5, 1 and 2 mg) were compared with placebo in a battery of tests of information processing, working and semantic memory. Peak saccadic velocity was used to provide a precise and reliable measure of sedation and its results were found to be consistent with those using visual analogue rating scales. Chlorpromazine 100 mg was equally sedative to lorazepam 2 mg. Lorazepam caused dose-dependent deterioration in performance in many of the memory tests, whereas an equally sedative dose of chlorpromazine did not. These data therefore support the view that benzodiazepine-induced amnesia is not secondary to sedation. Peak saccadic velocity has considerable advantages over visual analogue scales as a measure of sedation, since it is objective and has a demonstrated low coefficient of variation. It is suggested that saccadic eye movement measurement will permit considerably more reliable and precise separation of the sedative and amnestic effects of drugs and will allow investigation of amnesia caused by clinically relevant doses of psychotropic drugs.