Objective To assess the effectiveness and safety of progabide (PGB) for refractory epilepsy. Methods Randomized controlled trials (RCTs) on PGB treating refractory epilepsy were searched from the following databases as PubMed, EMbase, The Cochrane Library, CNKI, CBM and VIP from the date of their establishment to July 2011. The data of RCTs meeting the inclusive criteria were extracted according to Cochrane methods by two reviewers independently, and after the quality was evaluated and cross-checked, meta-analyses were conducted using RevMan 5.1 software. Results A total of seven studies involving 231 patients were included. The results of Meta-analyses showed that based on the conventional therapy, PGB was ineffective in treating refractory partial epilepsy compared with the placebo (OR=1.76, 95%CI 0.40 to 7.65, P=0.45), but it was superior to the placebo in treating refractory partial and generalized epilepsy (OR=4.46, 95%CI 2.06 to 9.65, P=0.000 1). The main adverse events of PGB were somnolence, dizziness and headache, which were mild and transient, which could turn to normal after reducing the dose of PGB and only a few patients needed to stop taking PGB. Conclusion Current studies shows that progabide may be effective in treating refractory partial and generalized epilepsy, but its effectiveness in treating refractory partial epilepsy is still unknown. The side effects of PGB are mostly mild. For the possibility of moderate selection bias existing in the quality of the included studies which may affect the authenticity of outcomes, so this conclusion still needs to be further proved by conducting more high-quality, large-scale and double-blinded RCTs.