MicroRNAs (miRNAs) are small non-coding RNAs which regulate gene expressions post-transcriptionally. Nowadays, various miRNAs have been found to be sensitive and specific biomarkers for the early diagnosis of colorectal cancer (CRC); however, there are different, even conflicting results in different publications concerning the diagnostic accuracy of miRNA. Therefore, we aim to conduct a meta-analysis of the relevant publications to comprehensively evaluate the diagnostic value of miRNAs in CRC detection. Several public databases such as PubMed, Embase, and Google Scholar were retrieved up to July 13, 2014. Sensitivity was applied to plot the summary receiver operator characteristic (SROC) curve against specificity. The area under the SROC curve (AUC) was calculated to assess the classified effects. STATA 12.0 software was used to perform all statistic analyses. A total of 29 articles, including 80 studies, were involved in our meta-analysis, 55 of which focus on single-miRNA assays and the other 25 on multiple-miRNA assays. Our results suggested that multiple-miRNA assays show a better diagnostic accuracy compared with single-miRNA assays. In addition, blood-based miRNA assays were more accurate than feces-based miRNA assays in CRC diagnosis. Our results also showed that miRNA diagnosis appear to be more accurate in Asians than in Caucasians. However, further researches are needed to validate our results and the feasibility of miRNAs as biomarkers in routine clinical diagnosis of CRC.