Abstract: Objective To explore the hypothesis that diabetes mellitus (DM) may cause hepatocellular carcinoma (HCC), by evaluating the association between DM and HCC by means of meta-analysis. Methods Cochrane Library，Medline，PubMed，EMBASE，OVID and CBMdisc databases were searched, and a manual search of citations from relevant original studies and literature also was performed. The data were pooled using RevMan 4.3 software for meta-analysis, and the potential bias was assessed by sensitivity analysis, funnel plot and fail-safe number. Results A total of 15 cohort studies were included, and 6 studies (the outcome of HCC was occurrence) were analyzed. DM was associated significantly with the occurrence of HCC (pooled RR=2.37, 95%CI:1.76~3.21). Sensitivity analysis showed the results of meta-analysis were insusceptible, publication bias was found in the studies, but it's effect was too limited to be ignored. Conclusion A significant causal association between DM and HCC was observed, it suggests that DM may be a risk factor of HCC.