BackgroundLiver cancer is a common malignancy with a high mortality rate. Given the poor prognosis associated with this cancer, many patients seek additional therapies that may improve quality of life or survival. Several Traditional Chinese Medicines (TCM) have been evaluated in clinical trials, but little is known about them outside of China.MethodsWe searched independently and in duplicate 8 electronic databases, including 2 Chinese language databases, until February 2009. We included any randomized clinical trials (RCT) evaluating a TCM oral preparation for the treatment of hepatocellular cancers. We abstracted data on survival, tumor response, and performance scores. We conducted a random-effects meta-analysis and applied a meta-regression analysis.ResultsWe included 45 RCTs (n = 3,236). All studies employed an active control group. In general, the reporting of methodological issues was poor. We analyzed data from 37 trials reporting on complete response effects score (Relative Risk [RR] of 1.26 (95 CI, 1.04–1.52, P = 0.01, I2 = 0%, P = 0.99). Products containing ginseng, astragalus and mylabris had a larger treatment effect (OR 1.34, 95% CI, 1.04–1.71, P = 0.01) than the pooled broad estimate, also the case for astragalus-based treatments (OR 1.35, 95% CI, 1.001–1.80. P = 0.048). We examined survival rates and pooled 15 studies reporting on 6 month outcomes (RR 1.10, 95% CI, 1.04–1.15, P = < 0.0001, I2 = 0%, P = 0.60). This effect was consistent at other prospective dates, including 12 months (22 trials, RR 1.26, 95% CI, 1.17–1.36, P = < 0.0001, I2 = 7%, P = 0.36), 24 months (15 trials, 1.72, 95% CI, 1.40–2.03, P = < 0.0001, I2 = 0%, P = 0.75); and, at 36 months (8 trials, RR 2.40, 95% CI, 1.65–3.49, P = < 0.0001, I2 = 0%, P = 0.62).LimitationsAll included trials were conducted in China where emerging evidence suggests many RCTs are not, in fact, randomized. Publication bias may exist, favouring positive reports.ConclusionOur meta-analysis displays compelling evidence of effectiveness for hepatocellular cancers that should be evaluated in high-quality and transparent clinical trials.