The aim of this study was to identify differentially-expressed miRNAs in the serum of non-small cell lung cancer (NSCLC) patients that might be a clinically-useful tool for lung cancer early detection. We performed miRNA expression profile analysis using TaqMan OpenArray Human panel in a discovery set of 70 serum samples obtained at lung tumor resection and 22 non-cancer subjects (NC). Selected serum miRNAs were then validated by quantitative PCR using an independent validation set of serum samples from LC patients (n = 84) and NC (n = 23). Sixty miRNAs were significantly up-regulated and 31 were down-regulated in the serum from NSCLC patients versus NC (adjusted p < 0.001). Four miRNAs (miR-193b, miR-301, miR-141 and miR-200b) were selected for validating their diagnostic value in an independent cohort. In the discovery set, the ROC plot derived from the combination of these miRNAs yielded an area under the curve (AUC) of 0.985 (95% CI 0.961-1.000, p < 0.001). In the test set, this miRNA signature exhibited an AUC of 0.993 (95% CI 0.979-1.000, p < 0.001). In conclusion, we identified a serum 4-miRNA signature that discriminated with high accuracy lung cancer patients from NC. Further prospective validation of this miRNA signature is warranted.