Alcohol misuse occurs commonly on college campuses, necessitating prevention programs to help college drinkers reduce consumption and minimize harmful consequences. Computer-delivered interventions (CDIs) have been widely used due to their low cost and ease of dissemination but whether CDIs are efficacious and whether they produce benefits equivalent to face-to-face interventions (FTFIs) remain unclear. Therefore, we identified controlled trials of both CDIs and FTFIs and used meta-analysis (a) to determine the relative efficacy of these two approaches and (b) to test predictors of intervention efficacy. We included studies examining FTFIs (N = 5,237; 56% female; 87% White) and CDIs (N = 32,243; 51% female; 81% White). Independent raters coded participant characteristics, design and methodological features, intervention content, and calculated weighted mean effect sizes using fixed and random-effects models. Analyses indicated that, compared to controls, FTFI participants drank less, drank less frequently, and reported fewer problems at short-term follow-up (d+s = 0.15 – 0.19); they continued to consume lower quantities at intermediate (d+ = 0.23) and long-term (d+ = 0.14) follow-ups. Compared to controls, CDI participants reported lower quantities, frequency, and peak intoxication at short-term follow-up (d+s = 0.13 – 0.29), but these effects were not maintained. Direct comparisons between FTFI and CDIs were infrequent, but these trials favored the FTFIs on both quantity and problems measures (d+s = 0.12–0.20). Moderator analyses identified participant and intervention characteristics that influence intervention efficacy. Overall, we conclude that FTFIs provide the most effective and enduring effects.