Many studies have demonstrated that epidermal growth factor receptor (EGFR) mutation is associated with the response to therapy with single agent EGFR-tyrosine kinase inhibitors (EGFR-TKI) in non-small cell lung cancer (NSCLC) patients, but the extent of the effect is varied. We carried out a meta-analysis to assess the association between EGFR mutation and the efficacy of EGFR-TKI therapy and the independent predictor of EGFR mutation in order to identify who would be likely to benefit from this kind of therapy. All literature relating to EGFR mutation and EGFR-TKI therapy was researched and carefully selected. Related variables were abstracted and the pooled odds ratio calculated after a heterogeneity test with the software, State 10. Publication bias was evaluated at the same time. Seventeen studies were included according to the selection criteria. We found a statistically significant higher probability of response in patients with an EGFR mutation versus wild-type (19.33, 95% CI, 13.61-27.46, P < 0.0001). Furthermore, the pooled odds ration of susceptibility to EGFR mutation among female patients compared with male patients was 3.01 (95% confidence interval (CI), 2.34-3.88 P < 0.0001), adenocarcinoma patients compared with non-adenocarcinoma patients was 5.40 (95% CI, 2.55-11.40 P < 0.0001), and non-smoker patients compared smoker patients was 19.33 (95% CI:,13.61-27.46 P < 0.0001). The publication bias analysis had no statistically significant results. Female, adenocarcinoma and non-smoker are independent predictors for the EGFR mutation. Efficacy of EGFR-TKI therapy favors patients with an EGFR mutation. Without the gene mutational analysis, patients selected for EGFR-TKI therapy should be female non-smokers with adenocarcinoma.