Objective: The aim of this study was to test whether the efficacy of in-person and computer delivered brief alcohol intervention (BAI) is moderated by mental health status. Method: General hospital inpatients with at-risk alcohol use aged 18 to 64 years (N = 961, 75% men) were allocated to in-person BAI, computer-based BAI, and assessment only. In-person BAI contained counseling by research staff. Computer-based BAI contained computer-generated individualized feedback letters. BAIs were designed to be delivered at baseline and 1 and 3 months later. Outcome was reduction in alcohol use per day after 6, 12, 18, and 24 months. Latent growth curve models were estimated. Two mental health indicators, the 5-item mental health inventory and routine care diagnosis of mental and behavioral disorders assessed by general hospital physicians, were tested as moderators of BAI efficacy. Results: In all groups, inpatients with better mental health reduced alcohol use after hospitalization (ps < 0.01). While inpatients with impaired mental health did not reduce their drinking significantly following assessment only, those who received any of the 2 BAIs did (ps < 0.05). Conclusions: BAI was particularly efficacious in reducing alcohol use among general hospital inpatients with at-risk alcohol use and impaired mental health, with computer-based delivery being at least as efficacious as in-person delivery.