Objectives: This meta-analysis of randomized controlled trials (RCTs) examined the efficacy and safety of minocycline for three major mental disorders: schizophrenia, bipolar disorder and major depressive disorder (MDD). Methods: A systematic literature search of major electronic databases was conducted. Meta-analysis of clinical efficacy as defined by the respective studies, all-cause discontinuation, adverse drug reactions (ADRs) with standardized mean difference (SMD) and risk ratios (RRs) and their 95% confidence intervals (CI) was conducted using random-effects model. Quality assessment was performed with the Jadad scale and Cochrane risk of bias. Results: Sixteen RCTs (n=1357) on minocycline (50–300 mg/day) for schizophrenia (13 RCTs, n=1196), bipolar depression (1 RCT, n=49), and MDD (2 RCTs, n=112) were analyzed separately by diagnosis. Twelve RCTs mentioned randomized allocation specifically; the weighted Jadad scores were 4.0. Adjunctive minocycline outperformed placebo in improving total psychopathology [SMD: −0.45 (95%CI: −0.73, −0.16), p=0.002; I=77%], positive [SMD: −0.15 (95%CI: −0.28, −0.02), p=0.02; I=0%], negative [SMD: −0.62 (95%CI: −0.95, −0.28), p=0.0003; I=85%] and general psychopathology scores [SMD: −0.28 (95%CI: −0.53, −0.03), p=0.03; I=59%] in schizophrenia. Minocycline showed no significant effect on depressive and manic symptoms in both bipolar depression and MDD. Minocycline caused significantly less headache (p=0.02, number-needed-to-harm=14, 95%CI=5–14) than placebo in schizophrenia. All-cause discontinuation and other ADRs were similar between minocycline and placebo in each diagnostic category. Conclusion: In this meta-analysis, adjunctive minocycline appeared to be efficacious and safe for schizophrenia. However, the efficacy of adjunctive minocycline for bipolar depression or MDD could not be demonstrated. Review registration: PROSPERO: CRD42018102483