ABSTRACT: Background: Epidermal growth factor receptor (EGFR) mutation testing in plasma cell-free DNA (cfDNA) from advanced lung cancer patients is an emerging clinical tool. This meta-analysis was designed to determine the diagnostic accuracy of two common PCR systems, droplet digital PCR (ddPCR) and amplification refractory mutation system PCR (ARMS-PCR), for detecting EGFR mutation in cfDNA. Materials and methods: A systematic search was carried out based on PubMed, Web of science, Embase and the Cochrane library. Data from eligible studies were extracted and pooled to calculate the sensitivity, specificity, diagnostic odds ratio (DOR), area under the summary receiver-operating characteristic curve (AUROC), using tissue biopsy results as the standard method. Subgroup analyses were performed regarding EGFR mutation type, tumor stage, and EGFR-TKI treatment. Results: Twenty-five studies involving 4,881 cases were included. The plasma testing sensitivity, specificity, DOR, and AUROC, compared with the matched tumor tissues, were 72.1%, 95.6%, 38.5, 0.89 for ddPCR, and 65.3%, 98.2%, 52.8, 0.71 for ARMS-PCR, respectively, through indirect comparison, significant differences were found in sensitivity (P = 0.003) and specificity (P = 0.007). Furthermore, significant difference was found in sensitivity between tumor stage subgroups (IIIB-IV subgroup vs. IA-IV subgroup) in ARMS-PCR (73.7 vs. 64.2%, P = 0.008), but not in ddPCR (72.5 vs. 71.2%, P = 0.756). Conclusions: This study demonstrates that ddPCR and ARMS-PCR have a high specificity with a practical sensitivity for detecting EGFR mutation in cfDNA, which supports their application as a supplement or a conditional-alternative to tissue biopsy in clinical practice for genotyping. It seems that ddPCR has a higher sensitivity than ARMS-PCR, especially in early stages.