BACKGROUND Alcohol use disorder (AUD) and alcohol-related impairments belong to the most widespread psychiatric disorders leading to specific psychophysical, affective and cognitive symptoms and consequences for psychosocial well-being and health. Alcohol consumption is increasingly becoming a problem in many developing regions and AUD prevalence is estimated at 4.1% worldwide, with highest prevalence in European countries (7.5%), and the North America (6.0%). Therapeutic approaches, including pharmacotherapy, play an important role in treating patients with AUD. OBJECTIVE To assess the efficacy and safety of baclofen for treating people with AUD, who are currently drinking, with the aim of achieving and maintaining abstinence or reducing alcohol consumption. METHODS We searched the Cochrane Drugs and Alcohol Specialised Register, CENTRAL, MEDLINE, Embase, two further databases and two clinical trials registries, conference proceedings, and the reference lists of retrieved articles. The date of the most recent search was 30 January 2018. METHODS Randomised controlled trials (RCTs) of at least four weeks' treatment duration and 12 weeks' overall study duration comparing baclofen for relapse prevention of AUD with placebo, no treatment or other treatments. METHODS We used standard methodological procedures expected by Cochrane. RESULTS We included 12 RCTs (1128 participants). All studies but three recruited fewer than 100 participants. Participants had a diagnosis of alcohol dependence according the Diagnostic and Statistical Manual of Mental Disorders (DSM) IV or the International Classification of Diseases (ICD)-10 criteria who were currently drinking. The mean age of participants was 48 years, and there were more men (69%), than women. All studies compared baclofen to placebo, except for one study that evaluated baclofen versus acamprosate. The included studies considered baclofen at different doses (range 10 mg a day to 150 mg a day). In all but one of the studies, participants in both the baclofen and placebo groups received psychosocial treatment or counselling of various intensity.We judged most of the studies at low risk of selection, performance, detection (subjective outcome), attrition and reporting bias.We did not find any difference between baclofen and placebo for the primary outcomes: relapse-return to any drinking (RR 0.88, 95% CI 0.74 to 1.04; 5 studies, 781 participants, moderate certainty evidence); frequency of use by percentage of days abstinent (MD 0.39, 95% CI -11.51 to 12.29; 6 studies, 465 participants, low certainty evidence) and frequency of use by percentage of heavy drinking days at the end of treatment (MD 0.25, 95% CI -1.25 to 1.76; 3 studies, 186 participants, moderate certainty evidence); number of participants with at least one adverse event (RR 1.04, 95% CI 0.99 to 1.10; 4 studies, 430 participants, high certainty evidence); the dropout rate at the end of treatment (RR 0.98, 95% CI 0.77 to 1.26, 8 studies, 977 participants, high certainty evidence) and dropout due to adverse events (RR 1.11, 95% CI 0.59 to 2.07; 7 studies, 913 participants, high certainty evidence).We found evidence that baclofen increases amount of use (drink per drinking days), (MD 1.55, 95% CI 1.32 to 1.77; 2 studies, 72 participants, low certainty evidence).Among secondary outcomes, there was no difference on craving (MD 1.38, 95% CI -1.28 to 4.03, 5 studies, 469 participants), and anxiety (SMD 0.07, 95% CI -0.14 to 0.28; 5 trials, 509 participants). We found that baclofen increased depression (SMD 0.27, 95% CI 0.05 to 0.48; 3 studies, 387 participants).Concerning the specific adverse events we found that baclofen increased: vertigo (RR 2.16, 95% CI 1.24 to 3.74; 7 studies, 858 participants), somnolence/sedation (RR 1.48, 95%CI 1.11 to 1.96; 8 studies, 946 participants), paraesthesia (RR 4.28, 95% CI 2.11 to 8.67; 4 studies, 593 participants), and muscle spasms/rigidity (RR 1.94, 95%CI 1.08 to 3.48; 3 studies, 551 participants). For all the other adverse events we did not find significant differences between baclofen and placebo.For the comparison baclofen versus acamprosate, we were only able to extract data for one outcome, craving. For this outcome, we found that baclofen increased craving compared with acamprosate (MD 14.62, 95% CI 12.72 to 16.52; 1 study, 49 participants). CONCLUSIONS None of the primary or secondary outcomes of the review showed evidence of a difference between baclofen and placebo. The high heterogeneity among primary studies results limits the interpretation of the summary estimate, the identification of moderators and mediators of baclofen's effects on alcohol use remains a challenge for further research. Even though some results from RCTs are promising, current evidence remains uncertain regarding the use of baclofen as a first-line treatment for people with AUDs.