OBJECTIVE To meta-analyze randomized controlled trials (RCTs) for the efficacy and safety of adjunctive antiepileptic drugs (AEDs) to augment clozapine therapy for treatment-resistant schizophrenia. METHODS The search included databases in English (PubMed, PsycINFO, Embase, and Cochrane Library databases and the Cochrane Controlled Trials Register) and in Chinese (China Journal Net [CJN], WanFang, and China Biology Medicine [CBM]) and references from retrieved articles. The databases were searched using dates inclusive from their onset until January 1, 2016, for terms reflecting (a) schizophrenia, (b) clozapine, and (c) adjunctive drugs. METHODS From 1,969 potentially relevant articles, 21 articles describing 22 RCTs were selected. METHODS Two independent investigators extracted data for a random-effects meta-analysis and assessed the quality of the studies using risk of bias and the Jadad scale. Standard mean difference, risk ratio (RR) ± 95% confidence intervals (CIs), and the number needed to harm (NNH) were used. RESULTS A total of 22 RCTs (N = 1,227) with 4 AEDs (topiramate [5 RCTs, n = 270], lamotrigine [8 RCTs, n = 299], sodium valproate [6 RCTs, n = 430], and magnesium valproate [3 RCTs, n = 228]) were analyzed. The means weighted by sample size were 12.1 weeks for treatment duration, 36.2 years for age, and 61% for male frequency. Significant superiority in total psychopathology was observed for topiramate (P < .0001), lamotrigine (P = .05), and sodium valproate (P = .002), compared to clozapine monotherapy. After removing outliers, the positive effect of sodium valproate remained, but the positive effect of lamotrigine disappeared (P = .40). Significantly improved efficacy in positive and general symptom severity was observed for topiramate (P = .04 and P = .02, respectively) and sodium valproate (P = .009 and P = .003, respectively). There were no significant differences regarding adverse drug reactions and all-cause discontinuations except for topiramate, which was associated with more all-cause discontinuations (RR = 1.99; 95% CI, 1.16 to 3.39; P = .01; I² = 0%; NNH = 7). CONCLUSIONS Sodium valproate augmentation was efficacious and safe. Topiramate augmentation had a too-high discontinuation rate. High-quality RCTs are needed to inform clinical recommendations.