BACKGROUND: Chronic antipsychotic drug treatment may cause tardive dyskinesia (TD), a long-term movement disorder. Gamma-aminobutyric acid (GABA) agonist drugs, which have intense sedative properties and may exacerbate psychotic symptoms, have been used to treat TD. OBJECTIVES: To determine the effects of GABA agonist drugs (baclofen, gamma-vinyl-GABA, gamma-acetylenic-GABA, progabide, muscimol, sodium valproate and tetrahydroisoxazolopyridine (THIP)) for people with antipsychotic-induced tardive dyskinesia (TD) and schizophrenia or other chronic mental illnesses. SEARCH STRATEGY: We updated the previous Cochrane review by searching the Cochrane Schizophrenia Group Register (September 2003). We searched references for further trial citations and, where possible, contacted authors. SELECTION CRITERIA: Randomised controlled trials comparing use of non-benzodiazepine GABA agonist drugs with placebo or no intervention, involving people with schizophrenia or other chronic mental illnesses with signs of antipsychotic-induced TD. DATA COLLECTION AND ANALYSIS: Working independently, we selected and critically appraised studies, extracted data and analysed on an intention-to-treat basis. Where possible and appropriate we calculated risk ratios (RR) and their 95% confidence intervals (CI) with the number needed to treat (NNT). For continuous data Weighted Mean Differences (WMD) were calculated. MAIN RESULTS: We identified eight small poorly reported studies for inclusion. For the outcome of 'no clinically important improvement in tardive dyskinesia' GABA agonist drugs were not clearly better than placebo (n = 108, RR 0.83 CI 0.6 to 1.1). Deterioration in mental state was more likely to occur in people receiving GABA medication (n = 95, RR 2.47 CI 1.1 to 5.4), but this effect was influenced by the decision to assign a negative outcome to those who dropped out before the end of the study. A greater proportion of people allocated GABA medication may fail to complete the trial compared with those allocated placebo (20% versus 9%), but this difference was not statistically significant (n = 136, RR 1.99 CI 0.8 to 4.7). There is a suggestion of an increase in ataxia (loss of power of muscular coordination) for both baclofen and sodium valproate (n = 95, RR 3.26 CI 0.4 to 30.2), and in sedation (n = 113, RR 2.12 CI 0.8 to 5.4) compared with placebo, but this was not significant. Withdrawal of tetrahydroisoxazolopyridine (THIP) may cause seizures. REVIEWERS' CONCLUSIONS: Evidence of the effects of baclofen, progabide, sodium valproate, or THIP for people with antipsychotic-induced TD is inconclusive and unconvincing. Any possible benefits are likely to be outweighed by the adverse effects associated with their use.