To date, many studies have shown that microRNAs (miRNA) exhibit altered expression levels in various cancers and may play a potential role as diagnostic and prognostic biomarkers of cancers. This meta-analysis was designed to evaluate the exact role of microRNA-31 (miR-31) for survival and discuss the possibility of utilizing miR-31 to predict the prognosis of patients with various human cancers. Electronic literature databases including PubMed, Web of Science, and Embase were searched for articles published until May 2014. The articles only written in English were considered. Data were extracted from studies comparing overall survival (OS), cancer-specific survival (CSS), or postoperative survival (PS) in patients with multiple cancers, which showed higher miR-31 expression than with similar patients. Pooled hazard ratios (HRs) of miR-31 for survival and 95 % confidence intervals (CI) were calculated. Ten studies with a total of 1,648 participants were included for the meta-analysis. For OS, the pooled HRs of higher miR-31 expression in cancers indicated significant predictor poorer survival in general cancers in either univariate analysis (HR=2.34, 95 % CI=1.15-3.52, P<0.05) or multivariate analysis (HR=1.15, 95 % CI=1.04-1.26, P<0.05). For CSS, elevated miR-31 was also a significant predictor to general cancers in multivariate analysis (HR=1.77, 95 % CI=1.06-2.47, P<0.05). And, no association was found between miR-31 expression and PS. In conclusion, the present findings indicate that high miR-31 expression is associated with poor OS and CSS in patients with general cancers and miR-31 may be a useful clinical prognostic biomarker.