The detection of epidermal growth factor receptor (EGFR) mutations in non-small cell lung cancer tissues is necessary for effective treatment with EGFR tyrosine kinase inhibitors. However, tumor tissues may not be available in all situations. Studies have evaluated the potential use of serum or plasma for detecting the EGFR mutation status, but the results have been inconclusive. Here, a meta-analysis was performed to determine whether blood samples could serve as substitutes for tissue specimens in detecting the EGFR mutation status. Databases, including PubMed and Embase, were searched for relevant studies published from 2005 to 2013 that included true-positive, false-positive, true-negative, and false-negative values of the EGFR mutation status of the blood compared with tissue specimens. Summary receiver operating characteristic curves were developed to explore the threshold effect. Spearman's correlation coefficient was calculated to analyze the heterogeneity between studies. Pooled sensitivity and specificity were evaluated using Meta-DiSc version 1.4. Thirteen articles involving 1,591 cases were enrolled, with a pooled sensitivity and specificity of 64.5 % (95 % CI = 0.605-0.683) and 88.5 % (95 % CI = 0.863-0.904), respectively. Heterogeneity among the studies was caused by factors other than threshold effect. The findings were influenced by test method (p = 0.0354). Blood samples had a high specificity and relatively low sensitivity for detecting EGFR mutations compared to tumor tissues. The results of this meta-analysis suggest that peripheral blood is insufficient as a substitute for tumor tissues in detecting EGFR mutations in clinical practice.