BACKGROUND AND AIMS: The proportion of untreated patients with alcohol use disorder (AUD) exceeds that of any other mental health disorder, and treatment alternatives are needed. A widely discussed strategy is to depart from the abstinence paradigm as part of controlled drinking approaches. This first systematic review with meta-analysis aims to assess the efficacy of non-abstinent treatment strategies compared with abstinence-based strategies. METHODS: CENTRAL, PubMed, PsycINFO and Embase databases were searched until February 2019 for controlled (randomized and non-randomized) clinical trials (RCTs and non-RCTs) among adult AUD populations, including an intervention group aiming at controlled drinking and a control group aiming for abstinence. Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and Cochrane Collaboration guidelines, literature search, data collection and risk of bias assessment were carried out independently by two reviewers [International Prospective Register of Systematic Reviews (PROSPERO), registration no. CRD42019128716]. The primary outcome was the proportion of participants consuming alcohol at or below the recommended threshold. Secondary outcomes were social functioning, drinking reductions, abstinence rates and dropouts. Using random-effects models, RCTs and non-RCTs were analyzed separately. Sensitivity and subgroup analyses accounted for methodological rigor, inclusion of goal-specific treatment, length of follow-up and AUD severity. RESULTS: Twenty-two studies (including five RCTs) with 4204 patients were selected. There was no statistically significant difference between both treatment paradigms in RCTs [odds ratio (OR) = 1.32, 95% confidence interval (CI) = 0.51-3.39]. Non-randomized studies of free goal choice favored abstinence-orientation (OR = 0.60, 95% CI = 0.40-0.90), unless goal-specific treatment was provided (OR = 0.79, 95% CI = 0.40-1.56), or in studies of low risk of bias (OR = 0.73, 95% CI = 0.49-1.09) or with long follow-up (OR = 1.49, 95% CI = 0.78-2.85). Effect sizes were not clearly dependent upon AUD severity. Abstinence- and controlled drinking interventions did not clearly differ in their effect on social functioning and drinking reductions. CONCLUSIONS: Available evidence does not support abstinence as the only approach in the treatment of alcohol use disorder. Controlled drinking, particularly if supported by specific psychotherapy, appears to be a viable option where an abstinence-oriented approach is not applicable.