The aims of this study were to (1) estimate the effect of extended-release naltrexone compared with placebo on alcohol consumption in patients with alcohol use disorder (AUD) and (2) conduct pre-planned subgroup analyses to test whether being abstinent when initiating treatment (lead-in abstinence) or the duration of treatment improves treatment efficacy. Systematic review and random-effects meta-analysis of blinded randomized placebo-controlled trials reporting the effect extended-release naltrexone on alcohol consumption. Outpatient clinics. Seven trials evaluating a total of 1500 adults with AUD receiving monthly injections of either placebo or extended-release naltrexone at doses of 150-400 mg for 2-6 months and some form of behavioral therapy. Pooled weighted mean difference (WMD) in drinking days per month and heavy drinking days per month. The WMD was -2.0 [95% confidence interval (CI) = -3.4, -0.6; P = 0.03] in favor of extended-release naltrexone for drinking days per month and -1.2 (95% CI = -0.2, -2.1; P = 0.02) for heavy drinking days per month, indicating that treatment resulted in two fewer drinking days per month and 1.2 fewer heavy drinking days per month compared with placebo. Trials not requiring lead-in abstinence and those lasting longer than 3 months reported larger reductions in heavy drinking days per month; WMD -2.0 (95% CI = -3.52, -0.48; P = 0.01) and -1.9 (95% CI = -3.2, -0.5; P = 0.01), respectively. In all cases, the I2 statistics (0-7.2%) did not suggest substantial heterogeneity. Extended-release naltrexone reduces drinking days and heavy drinking days per month compared with placebo. Reductions are larger with a longer duration of treatment.