Treatment of primary biliary cirrhosis is complicated. There are studies suggesting that bezafibrate, alone or in combination with ursodeoxycholic acid (UDCA), is effective in the treatment of primary biliary cirrhosis, but no systematic review has summarised the evidence yet.To assess the beneficial and harmful effects of bezafibrate in patients with primary biliary cirrhosis.The Cochrane Hepato-Biliary Group Controlled Trials Register, The Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library, MEDLINE, EMBASE, Science Citation Index Expanded, LILACS, Clinicaltrials.gov, the WHO International Clinical Trials Registry Platform, and full text searches were conducted until November 2011. The searches in Chinese Bio-medical Literature Database, China Network Knowledge Information, Chinese Science Journal Database, Chinese Medical Citation Index, Wanfang Database, and full text searches were conducted until January 2011. Manufacturers and authors were contacted.All randomised clinical trials comparing bezafibrate at any dose or regimen in patients with primary biliary cirrhosis with placebo or no intervention, or with another drug. Any concomitant interventions were allowed if received equally by all treatment groups in a trial.Two authors extracted data. RevMan Analysis was used for statistical analysis of dichotomous data with risk ratio (RR) or risk difference (RD), and of continuous data with mean difference (MD), both with 95% confidence intervals (CI). Methodological domains were used to assess risk of systematic errors (bias). Trial sequential analysis was used to control for random errors (play of chance).Six trials with 151 Japanese patients were included. All trials had high risk of bias. Four trials compared bezafibrate plus UDCA with no intervention plus UDCA (referenced as bezafibrate versus no intervention in the remaining text), and two trials compared bezafibrate with UDCA. No patient died and no patient developed liver-related complications in any of the included trials. Bezafibrate was without significant effects on the occurrence of adverse events compared with no intervention (5/32 (16%) versus 0/28 (0%)) (RR 5.40, 95% CI 0.69 to 42.32; 3 trials with 60 patients; I² = 0%) or with UDCA (2/32 (6%) versus 0/37 (0%)) (RR 6.19, 95% CI 0.31 to 122.05; 2 trials with 69 patients; I² = 0%). Bezafibrate significantly decreased the activity of serum alkaline phosphatases compared with no intervention (MD -186.04 U/L, 95% CI -249.03 to -123.04; 4 trials with 79 patients; I² = 34%) and when compared with UDCA (MD -162.90 U/L, 95% CI -199.68 to -126.12; 2 trials with 48 patients; I² = 0%). These results were supported by trial sequential analyses. Bezafibrate compared with no intervention significantly decreased plasma immunoglobulin M (MD -164.00 mg/dl, 95% CI -259.47 to -68.53; 3 trials with 50 patients; I² = 46%) and serum bilirubin concentration (MD -0.19 mg/dl, 95% CI -0.38 to -0.00; 2 trials with 34 patients; I² = 0%). However, the latter two results were not supported by trial sequential analyses. Bezafibrate compared with no intervention had no significant effect on the activity of serum gamma-glutamyltransferase (MD -1.22 U/L, 95% CI -11.97 to 9.52; 4 trials with 79 patients; I² = 42%) and serum alanine aminotransferase (MD -5.61 U/L, 95% CI -24.50 to 13.27; 2 trials with 35 patients; I² = 34%). Bezafibrate compared with UDCA had no significant effect on the activity of serum gamma-glutamyltransferase (MD 38.44 U/L, 95% CI -180.67 to 257.55; 2 trials with 49 patients; I² = 89%), serum alanine aminotransferase (MD -2.34 U/L, 95% CI -34.73 to 30.06; 2 trials with 49 patients; I² = 95%), and plasma immunoglobulin M concentration (MD -20.23 mg/dl, 95% CI -218.71 to 178.25; 2 trials with 41 patients; I² = 90%) in random-effects model meta-analyses, but bezafibrate significantly decreased the activity of serum gamma-glutamyltransferase (MD -58.18, 95% CI -76.49 to -39.88; 2 trials with 49 patients; I² = 89%), serum alanine aminotransferase (MD -13.94, 95% CI -18.78 to -9.09; 2 trials with 49 patients; I² = 95%), and plasma immunoglobulin M concentration (MD -99.90, 95% CI -130.72 to -69.07; 2 trials with 41 patients; I² = 90%) in fixed-effect model meta-analyses. One patient had bezafibrate withdrawn due to an adverse event compared to no intervention (RD 0.03, 95% CI -0.09 to 0.16; 2 trials with 60 patients; I² = 0%).This systematic review did not demonstrate any effect of bezafibrate versus no intervention on mortality, liver-related morbidity, adverse events, and pruritus in patients with primary biliary cirrhosis. Furthermore, we found no significant effects of bezafibrate on mortality, liver-related morbidity, or adverse events when compared with ursodeoxycholic acid, None of the trials assessed quality of life or fatigue. The data seem to indicate a possible positive intervention effect of bezafibrate on some liver biochemistry measures compared with the control group, but the observed effects could be due to systematic errors or random errors. We need more randomised clinical trials on the effects of bezafibrate on primary biliary cirrhosis with low risks of systematic errors and random errors.