ABSTRACT Objective To evaluate moderators and mediators of brief alcohol interventions conducted in the emergency department. Methods Patients (18–24 years; n = 172) in an emergency department received a motivational interview with personalized feedback (MI) or feedback only (FO), with 1- and 3-month booster sessions and 6- and 12-month follow-ups. Gender, alcohol status/severity group [ALC+ only, Alcohol Use Disorders Identification Test (AUDIT+) only, ALC+/AUDIT+], attribution of alcohol in the medical event, aversiveness of the event, perceived seriousness of the event and baseline readiness to change alcohol use were evaluated as moderators of intervention efficacy. Readiness to change also was evaluated as a mediator of intervention efficacy, as were perceived risks/benefits of alcohol use, self-efficacy and alcohol treatment seeking. Results Alcohol status, attribution and readiness moderated intervention effects such that patients who had not been drinking prior to their medical event, those who had low or medium attribution for alcohol in the event and those who had low or medium readiness to change showed lower alcohol use 12 months after receiving MI compared to FO. In the AUDIT+ only group those who received MI showed lower rates of alcohol-related injury at follow-up than those who received FO. Patients who had been drinking prior to their precipitating event did not show different outcomes in the two interventions, regardless of AUDIT status. Gender did not moderate intervention efficacy and no significant mediation was found. Conclusions Findings may help practitioners target patients for whom brief interventions will be most effective. More research is needed to understand how brief interventions transmit their effects.