This study examined the separate and combined effects of alcohol (0.0 or 0.5 g/kg) and alcohol expectancies (none or 2-3 standard drinks) on risk-taking using a simulated-driving lane choice task. In this task, risk-taking was operationalized as choosing a cone-defined lane with a higher relative probability of hitting a cone. When alcohol was received but not expected, the probability of a risky lane choice increased compared with when alcohol was neither expected nor received. However, when subjects both expected and received alcohol, the probability of a risky lane choice was significantly decreased compared with when alcohol was neither expected nor received. These findings suggest that the knowledge of dose received can differentially influence the pharmacological effect of alcohol on decision-making.