Alcohol attitudes predict unique variance in drinking behavior and have been the target of manipulations and interventions to reduce high-risk alcohol use among youth and adults. However, whether these manipulations create long-lasting changes in alcohol-related attitudes and drinking behavior is unclear. The current mini-review focuses on evaluative conditioning (EC), a manipulation which pairs alcohol-related stimuli repeatedly with affectively valanced stimuli to create new semantic associations in memory; such associations underlie reflexive or impulsive behaviors like high-risk alcohol use. Across experimental studies, EC has been shown to promote negative alcohol attitudes and reduce alcohol consumption. However, recent evidence suggests the effectiveness of EC may depend on the depth of learning facilitated during the task, which may strengthen the semantic associations through propositional learning. While researchers have experimentally promoted greater depth of learning through the manipulation of contextual factors, we review evidence that alcohol-related individual differences also impact the effectiveness of alcohol EC, particularly when these factors are explicitly linked to the stimuli used during the manipulation. This review provides future directions for researchers and practitioners aiming to shape alcohol-related attitudes and behaviors. Specifically, the malleability of alcohol-related attitudes may depend on propositional learning facilitated by contextual and individual factors. Researchers and practitioners should incorporate these factors into interventions like EC aiming to reduce high-risk alcohol consumption.