OBJECTIVE: Previous studies suggested that metformin is associated with decreased risk of cancer; however, results specifically addressing the potential association with prostate cancer were limited and contradictory. This study considers the association between metformin and the incidence, mortality and recurrence of prostate cancer by performing a meta-analysis of observational studies. METHODS: Literatures published before January 2014 were searched by using databases of PubMed and Embase. Pooled relative risks (RRs) were determined using a random effects model to evaluate the strength of association between metformin therapy and risk of prostate cancer. RESULTS: Thirteen studies involving a total of 334 532 participants were included in this meta-analysis. Compared with the control group, metformin therapy was associated with significantly decreased incidence of prostate cancer [RR = 0.88, 95% confidence interval (CI) [0.78, 0.99], p = 0.03, I(2)  = 74.7%]. However, metformin therapy was not associated with decreased all-cause mortality (RR = 1.07, 95% CI [0.86, 1.32], p = 0.55, I(2)  = 58.2%) or decreased recurrence of prostate cancer (RR = 0.90, 95% CI [0.75, 1.09], p = 0.27, I(2)  = 0.0%). No publication bias was detected (pBegg  = 0.55, pEgger  = 0.46). CONCLUSIONS: The present study suggested that metformin therapy may decrease the incidence of prostate cancer but that there was no association between the treatment and all-cause mortality or recurrence. It is recommended that this finding should be considered carefully and confirmed with further studies.