Three drinks containing 0 g (no alcohol, NA), 8 g (low alcohol, LA) and 24 g (high alcohol, HA) of alcohol were formulated which were found to be indistinguishable from one another in sip-and-swallow triangle tests. In a second study, conducted according to a within-subjects design, 14 healthy human volunteers consumed these drinks as part of a small lunchtime meal, in counterbalanced order on 3 different days. They also completed a battery of cognitive tasks, together with mood ratings, before lunch and during the 4 h following lunch. Compared with NA, LA (approximately 0.12 g/kg) significantly increased hit rate on a difficult rapid information processing vigilance task. In contrast, HA (approximately 0.35 g/kg) tended to impair performance of this task. There were no reliable effects of alcohol on performance on less demanding tasks. The low dose of alcohol also improved mood (for example, it significantly reduced tension and uncertainty), suggesting that the improvement in task performance was mediated by the calming or sedative effects of the alcohol. Volunteers did detect alcohol in the HA, but not the LA drink, when they consumed the full drink, confirming the difficulty of disguising the administration of alcohol.