This meta-analysis examined the effectiveness and safety of metformin to prevent or treat weight gain and metabolic abnormalities associated with antipsychotic drugs. We systematically searched in both English- and Chinese-language databases for metformin randomized controlled clinical trials (RCTs) using placebo in patients taking antipsychotics. Twenty-one RCTs (11 published in English and 10 in Chinese) involving 1547 subjects (778 on metformin, 769 on placebo) were included in this meta-analysis. Metformin was significantly superior to placebo (standard mean differences, −0.69 to −0.51; P = 0.01–0.0001) in the primary outcome measures (body weight, body mass index, fasting glucose, fasting insulin, triglycerides, and total cholesterol). Metformin was significantly superior to placebo in some secondary outcome measures but not in others. Significantly higher frequencies of nausea/vomiting and diarrhea were found in the metformin group, but no differences were found in other adverse drug reactions. In the metformin group, the frequency of nausea/vomiting was 14%, and of diarrhea, 7%. Subgroup and sensitivity analyses demonstrated that primary outcomes were influenced by ethnicity, treatment style (intervention vs prevention), metformin dose, study duration, and mean age. Body weight standard mean difference was −0.91 (confidence interval [CI], −1.40 to −0.41) in 3 prevention RCTs in naive patients, −0.66 (CI, −1.02 to −0.30) in 5 intervention RCTs during the first year, and −0.50 (CI, −0.73 to −0.27) in 9 intervention RCTs in chronic patients. This meta-analysis suggests that adjunctive metformin is an effective, safe, and reasonable choice for antipsychotic-induced weight gain and metabolic abnormalities.