Cancer has been a major public health issue all over the world and cancer patients diagnosed at early stages have a comparatively favorable prognosis. The association between specific dysregulated expressed microRNA-155 (miRNA-155, miR-155) and tumorigenesis has been identified by numerous studies. However, perplexity and inconsistence arise from a wide range of studies due to heterogeneity. Therefore, this meta-analysis was carried out to validate the association between miR-155 and tumorigenesis together with the clinical applicability of miR-155.Relevant studies were searched, identified, and selected from PubMed, Embase, Cochrane, Sinomed, and Wanfang database until July 5, 2015. Then, the sensitivity, specificity, and area under the summary receiver operator characteristic curve (AUC) were calculated to assess the overall performance miR-155 for cancer detection.A total of 25 studies were included in the meta-analysis with a total number of 1896 cancer patients and 1226 healthy controls. The overall sensitivity and specificity was 76.8% (95%CI: 71.1-81.7%) and 82.9% (95% CI: 77.5-87.3%), respectively. In addition, the pooled AUC and partial AUC was 0.867 and 0.718, respectively. Results from subgroup analyses suggested that the diagnostic accuracy of miR-155 in the Caucasian group was significantly higher than that in the Asian group. Similarly, serum sample type may provide better diagnostic value of miR-155 than plasma. Apart from that, miR-155 in breast cancer achieved the highest accuracy compared with miR-155 in other types of cancer.Results from meta-analysis suggested that miR-155 had great potential as a novel noninvasive biomarker for human cancer detection, particularly when breast cancer or Caucasian is involved. However, well-designed cohort or case control studies with large sample size should be implemented to confirm the diagnostic value of miR-155.