Forty-seven at-risk drinkers attended two sessions. On the first lab session, participants underwent one of two prediction error-generating procedures in which outcome expectancies were violated while retrieving alcohol memories (omission and value prediction error groups). Participants in a control group retrieved non-alcohol memories. Participants then reappraised personally relevant maladaptive alcohol memories and completed measures of reappraisal recall, alcohol verbal fluency and craving. Seven days later, they repeated these measures along with attentional bias assessment.