Introduction: Lung cancer is the deadliest cancer worldwide. Lung cancer screening has been applied in high-risk subjects and has proven to reduce lung cancer deaths in high-income countries. Biomarkers detection in body fluids may guide more precise selection of high-risk subjects but this approach has not been currently employed for lung cancer screening. Aims: To identify a miRNA-based signature for early detection of lung cancer in liquid biopsies. Methods: We analyzed plasma and sputum samples from high-risk subjects (n=54), and patients diagnosed with non-small cell lung cancer (n=60). Samples were subjected to RNA isolation followed by miRNA expression using Human v3 miRNA panel (NanoString technologies). Counts were normalized by housekeeping candidates, and differentially expressed miRNAs were filtered out by Rosalind software according to fold-change (?|1.3|), p-value (?0.01), and AUC>0.70. Results: Overall, high-risk subjects were stratified in LungRADS1, and LungRADS2, and cancer patients were diagnosed with adenocarcinoma, squamous cell carcinoma, and others. No differences were observed related to age, sex, and tobacco exposure in high-risk and cancer groups (p>0.05). For plasma, we observed seven differentially expressed miRNAs able to distinguish cancer from non-cancer samples, and three of them presented with high accuracy. For sputum, we observed three differentially expressed miRNAs to distinguish cancer from non-cancer samples, and two of them presented with high accuracy. Conclusion: We identified fluid-specific miRNA signatures in liquid biopsies potentially to be employed in lung cancer screening programs to better guide selection of high-risk subjects for early detection of lung cancer.