Two experiments investigated the effect of alcohol on retrieval of lexical information. In each, volunteers received alcohol (1 ml per kg body weight) in one session and no alcohol in another in counterbalanced order. Experiment 1 was a computerised version of the Mill Hill vocabulary test in which subjects were required to define words by making multiple choice responses as fast as possible. As expected, correct decision time increased with item difficulty and tended to increase with alcohol, but there was no interaction between these effects. Experiment 2 was a lexical decision task involving words of low, medium and high frequency. Alcohol significantly increased correct response time but this did not interact with word frequency. In both experiments, decision times for individual items varied, indicating that lexical access is more difficult for rare than for frequent items. However, alcohol slowed easy and difficult decisions equally, which suggests that its locus of effect is not primarily on speed of access to semantic information, but rather on other aspects of the decision process.