OBJECTIVE: The purpose of this study was to test whether the link between alcohol expectancies and alcohol use (drinking frequency, usual quantity, five-plus drinking) is mediated by drinking motives. METHOD: Linear structural equation models were estimated based on a nationally representative sample of 5,616 8th, 9th, and 10th graders in Switzerland (51% female; mean [SD] age = 15.1 [1.0] years). RESULTS: In most cases, a perfect mediation occurred. Although all expectancy and motive dimensions were related to all alcohol-use measures in multivariate models, the expectancy link in multiple multivariate models was reduced to zero, whereas the motive link remained basically the same. One exception was the Tension Reduction Expectancy scale, which included aspects other than problem coping that were still related to alcohol consumption, even when coping motives were controlled for. CONCLUSIONS: Given the consistency of the results across different alcohol expectancies, drinking motives, and alcohol-use measures, the present study provides evidence to support one basic assumption of the motivational model of alcohol use: Drinking motives are the most proximate factor that precedes alcohol use. They are the gateway through which more distal influences (e.g., alcohol expectancies) are mediated.