Rationale. The alcohol discriminative stimulus has been extensively studied in animals and demonstrated to be pharmacologically complex. In contrast, however, the alcohol stimulus has been less frequently studied in humans. Objectives. The aim of the experiments reported here was to characterise pharmacologically an alcohol discriminative stimulus in social drinkers. Methods. Volunteers were first trained to discriminate a dose of 0.2 g/kg alcohol from placebo, using an established method. We then investigated the generalisation response and subjective effects following a range of doses of the %-amino-butyric acid (GABA)A benzodiazepine-receptor agonist lorazepam (0, 0.5, 1 and 2 mg, PO). Results. Low doses of lorazepam (0.5 and 1 mg) did not cross-generalise with the alcohol stimulus and produced only minimal subjective effects. However, a dose of 2 mg lorazepam substituted (60.8%) for the stimulus (P<0.02) and increased subjective ratings of "lightheaded" (P<0.05). Conclusions. These results are consistent with the pre-clinical literature and indicate the cross-species generality of the GABAA component of the alcohol discriminative stimulus.