Objective: Studies have shown that expectations of alcohol-induced impairment can produce adaptive responses to alcohol that reduce the degree of behavioral impairment displayed. The present study tested psychomotor performance following combined caffeine and alcohol administration in 42 social drinkers (23 men). Subjects were led to expect either that caffeine would antagonize alcohol-induced impairment or that it would have no effect. The study tested the hypothesis that drinkers who expected an antagonist effect of caffeine would display greater alcohol impairment than those who expected no antagonist effect. Method: Groups practiced a pursuit rotor task and received a moderate dose of alcohol (0.65 g/kg) combined with either 4.0 mg/kg caffeine or placebo caffeine. Some groups were led to expect that caffeine would counteract the impairing effect of alcohol and others were led to expect no counteracting effect. Psychomotor performance was then tested over a 3-hour period. Results: In accord with the hypothesis, groups led to expect counteracting effects of caffeine displayed greater impairment than those led to expect no counteraction. Caffeine had no significant antagonist effect on alcohol impairment. Conclusions: The findings suggest that compensation for alcohol impairment occurs when drinkers hold clear expectations that the drug will disrupt performance. When no such clear expectation exists, no compensatory response occurs and the impairing effects of alcohol are observed.