BACKGROUND: Epilepsy is one of the most common neurological disorders. Many people with epilepsy are drug‐resistant and require add‐on therapy, meaning that they concomitantly take multiple antiepileptic drugs. Carisbamate is a drug which is taken orally and inhibits voltage‐gated sodium channels. Carisbamate may be useful for drug‐resistant focal epilepsy. OBJECTIVES: To evaluate the efficacy and tolerability of carisbamate when used as an add‐on therapy for drug‐resistant focal epilepsy. SEARCH METHODS: We searched the following databases on 8 April 2021: Cochrane Register of Studies (CRS Web) and MEDLINE (Ovid) 1946 to April 07, 2021. CRS Web includes randomised or quasi‐randomised controlled trials from PubMed, Embase, ClinicalTrials.gov, WHO ICTRP, the Cochrane Central Register of Controlled Trials (CENTRAL), and the specialised registers of Cochrane review groups including Epilepsy. We also searched ongoing trials registers, checked reference lists, and contacted authors of the included trials. SELECTION CRITERIA: Double‐blind randomised controlled trials (RCTs) comparing carisbamate versus placebo or another antiepileptic drug, as add‐on therapy for drug‐resistant focal epilepsy. Trials could have a parallel‐group or cross‐over design. DATA COLLECTION AND ANALYSIS: Two review authors independently selected the trials for inclusion, assessed trial quality, and extracted data. The primary outcome was 50% or greater reduction in seizure frequency (responder rate). The secondary outcomes were: seizure freedom, treatment withdrawal (for any reason and due to adverse events); adverse events, and quality of life. We analysed data using the Mantel‐Haenszel statistical method and according to the intention‐to‐treat population. We presented results as risk ratios (RRs) with 95% confidence intervals (CIs). MAIN RESULTS: We included four RCTs involving a total of 2211 participants. All four trials compared carisbamate with placebo for drug‐resistant focal epilepsy. Participants in all trials were over 16 years of age and received at least one other antiepileptic drug concomitantly. We detected substantial risk of bias across the included trials. All four trials were at high risk of attrition bias due to the incomplete reporting of attrition and the high treatment withdrawal rates noted, especially with higher doses. All four trials also had unclear risk of detection bias, as they did not specify whether outcome assessors were blinded. Meta‐analysis suggested that carisbamate produced a higher responder rate compared to placebo (RR 1.36, 95% CI 1.14 to 1.62; 4 studies; moderate‐certainty evidence). More participants in the carsibamate group achieved seizure freedom (RR 2.43, 95% CI 0.84 to 7.03; 1 study); withdrew from treatment for any reason (RR 1.32, 95% CI 0.82 to 2.12; 4 studies); and withdrew from treatment due to adverse events (RR 1.80, 95% CI 0.78 to 4.17; 4 studies) than in the placebo group. However, the evidence for the three outcomes was very low‐certainty. There was no difference between treatment groups for the proportion of participants experiencing at least one adverse event (RR 1.10, 95% CI 0.93 to 1.30; 2 studies; low‐certainty evidence). More participants in the carisbamate group than in the placebo group developed dizziness (RR 2.06, 95% CI 1.23 to 3.44; 4 studies; very low‐certainty evidence) and somnolence (RR 1.82, 95% CI 1.28 to 2.58; 4 studies; low‐certainty evidence), but not fatigue (RR 1.11, 95% CI 0.73 to 1.68; 3 studies); headache (RR 1.13, 95% CI 0.92 to 1.38; 4 studies); or nausea (RR 1.19, 95% CI 0.81 to 1.75; 3 studies). None of the included trials reported quality of life. AUTHORS' CONCLUSIONS: The results suggest that carisbamate may demonstrate efficacy and tolerability as an add‐on therapy for drug‐resistant focal epilepsy. Importantly, the evidence for all outcomes except responder rate was of low to very low certainty, therefore we are uncertain of the accuracy of the reported effects. The certainty of the evidence is limited by the significant risk of bias associated with the included studies, as well as the statistical heterogeneity detected for some outcomes. Consequently, it is difficult for these findings to inform clinical practice. The studies were all of short duration and only included adult study populations. There is a need for further RCTs with more clear methodology, long‐term follow‐up, more clinical outcomes, more seizure types, and a broader range of participants.