Background: MicroRNA is a class of small, well-conserved, non-coding RNAs, and could play a potential role as diagnostic and prognostic biomarkers of esophageal cancers. We aimed to review comprehensively the evidence of microRNA as prognostic biomarkers in esophageal cancers. Methods: Studies were identified by searching PubMed, Embase and Web of Science until November 2013. Descriptive characteristics of studies were described and an additional meta-analysis for specific microRNAs which were studied most frequently was performed. Pooled hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs) were calculated. Fixed model or random model method was chosen depending on the heterogeneity among the studies. Results: Twenty-two studies including a total of 1946 participants were enrolled after a strict filtering and qualifying process. Among 33 prognostic microRNAs identified for esophageal cancer, miR-21 and miR-375 appeared more frequently. The median study size was 70.5 patients (29-249 patients) and the median HR was 3.305 (IQR = 1.615-7.31). For the studies evaluating miR-21's association with overall survival (OS), the pooled HR suggested that high level of miR-21 has a negative impact on OS (HR= 1.52[1.17-1.981, P= 0.001). As for miR-375, the pooled HR for OS (high/low) was 0.53 (95% CI: 0.39-0.73, P < 0.001), indicated that low level of miR-375 has a negative impact on OS. These results indicated that microRNAs show promising associations with prognosis in esophageal cancer. Up-regulation of miR-21 and down-regulation of miR-375 can predict unfavourable prognosis in esophageal cancer. (C) 2014 Elsevier Inc. All rights reserved.