Recent studies have indicated that long non-coding RNAs (lncRNAs) could act as non-invasive tumor markers in both diagnosis and predicting the prognosis. In this study, we focused to determine the expression of circulating lncRNAs in patients suffering from non-small-cell lung cancer (NSCLC), aiming to found the potential lncRNA as predictor. Twenty-one lncRNAs which previously identified were selected as candidate targets for subsequent circulating lncRNA assay. The candidate lncRNAs were validated by qRT-PCR arranged in the training and validation sets. Circulating SPRY4-IT1, ANRIL, and NEAT1 were significantly increased in plasma samples of NSCLC patients during training set and validation set. Receiver operating characteristic curve (ROC) analysis revealed that plasma ANRIL provided the highest diagnostic performance with an area under ROC curve value (AUC) of 0.798. Further combination with the three factors indicated a higher power (AUC, 0.876; sensitivity, 82.8 %; specificity, 92.3 %). The stableness detection of the three factors indicated that circulating SPRY4-IT1, ANRIL, and NEAT might serve as a predictor for the early warning of non-small-cell lung cancer.