Currently, the combination of ultrasonography and fine-needle aspiration biopsy (FNAB) can not discriminate between benign and malignant tumor of thyroid in some cases. The main issue in assessing the patients with thyroid nodules is to distinguish thyroid cancer from benign nodules, and reduce diagnostic surgery. To identify potential molecular biomarkers for patients with indeterminate FNAB, we explored the differentially expressed genes (DEGs) and differentially expressed long non-coding RNAs (DElncRNAs) in TCGA database between 318 papillary thyroid carcinoma (PTC) tissues and 35 normal thyroid gland tissues by DESeq R. Furthermore, DEGs were verified by gene expression profile GSE33630. Ten top DEGs and DElncRNAs were identified as candidate biomarkers for diagnosis and Lasso (Least Absolute Shrinkage and Selection Operator) logistic regression analysis were performed to improve the diagnostic accuracy of them. Besides, partial molecular biomarkers of top DEGs and DElncRNAs were closely related to the tumor stage (T), lymph node metastasis (N), metastasis (M) and pathological stage of PTC, which could reflect behavior of tumor progression. According to multivariate Cox analysis, the combination of two DEGs (METTL7B and KCTD16) and two DElncRNAs (LINC02454 and LINC02471) could predict the outcome in a more exact way. In conclusion, top DEGs and DElncRNAs could raise diagnosis of PTC in indeterminate FNAB specimens, and some could function as molecule biomarkers for tumor progression and prognosis.