Two key aspects of alcohol expectancy theory--(a) that memories about alcohol effects are stored as relatively cohesive templates of information and (b) that these templates are automatically activated in alcohol-related contexts--were tested using the Deese-Roediger- McDermott false memory paradigm. Alcohol expectancy adjectives were studied, and false memory for expectancy target words was tested in neutral and alcohol contexts. Results indicated that in the alcohol context heavier drinkers showed more false memory for alcohol expectancy words than they did in a neutral context. Differences were not found for lighter drinkers. These results were consistent with alcohol expectancy theory, which was then compared with various forms of association theory in explaining these results and larger issues in the addiction field.