Evidences from randomized controlled trials (RCTs) for the efficiency of traditional Chinese medicine (TCM) on the treatment of nonalcoholic fatty liver disease (NAFLD) are conflicting. Here we conducted a systematic review and meta-analysis of RCTs to evaluate the efficiency and safety of TCM in the treatment of NAFLD. Studies were searched on PubMed and China National Knowledge Infrastructure from January 1995 to June 2010. RCTs comparing either TCM formulations alone or in combination with placebo, ursodeoxycholic acid, insulin sensitizers, lipid-lowering drugs, or antioxidants were included. The category of most usually used herbs in the treatment of NAFLD was also calculated. Five thousand nine hundred and four patients from 62 RCTs were included for meta-analysis and 25,661 patients from 419 clinical studies were for TCM formulation analysis. Comparing with western medicines mentioned above, TCM had a better effect on the normalization of alanine aminotransferase and disappearance of radiological steatosis in the treatment of NAFLD. Furthermore, 246 kinds of Chinese herbs were included in our present study, with an average of 10 herbs (range 1-31) in each formulation. Hawthorn Fruit (321 times in 17,670 patients) was the most often used herb in the treatment of NAFLD. In conclusion, TCM is of modest benefit to the treatment of NAFLD.