As a psychostimulant, caffeine is thought to reduce road accidents by keeping drivers alert and wakeful. Studies have found that caffeine can improve performance on vigilance tasks and in driving simulators under normal sleeping conditions and after sleep restriction or deprivation. However, there is increasing evidence that these beneficial effects of caffeine are due to withdrawal reversal. Studies comparing the effects of caffeine versus placebo on driving performance have tested habitual caffeine consumers deprived of caffeine from the evening before the test day. The conclusion from this review is, therefore, that improvements in driving performance and alertness after caffeine are likely to represent withdrawal reversal rather than a net beneficial effect of caffeine. Further research using designs that control for caffeine withdrawal are necessary and, accordingly, advice given to the public on use of caffeine as an antidote to tiredness and impaired performance should be reviewed.