Thyroid nodules are common in the general population and vary widely in their propensity to harbor thyroid malignancies. The category of follicular lesion of undetermined significance, for instance, carries only a 15% risk of malignancy. The overarching aim of this work was the proteomic study of thyroid cancer because more effort needs to be placed on differentiating malignant thyroid nodules to avoid unnecessary thyroidectomy. We used 2-dimensional electrophoresis coupled to nano-liquid chromatography electrospray ionization tandem mass spectrometry, to examine fine-needle aspiration (FNA), which was easily attainable from the wash of the syringe used for classical FNA biopsy. Overall, we found 25 different proteins able to discriminate benign from malignant samples. The different expression of moesin; annexin A1 (ANXA1); cornulin (CRNN); lactate dehydrogenase; enolase; protein DJ-1; and superoxide dismutase was confirmed in FNA by enzyme-linked immunosorbent assay or Western blot. Receiver operating characteristic curves were calculated to investigate the discriminative power of our marker. The best performance in diagnosis was obtained by combining ANXA1, enolase, protein DJ-1, superoxide dismutase, and CRNN. In addition, the most highly ranked proteins, from the perspective of follicular lesion of undetermined significance, were ANXA1 and CRNN. The research of these candidate biomarkers has then been widened to other biological fluids, such as serum and whole saliva. In conclusion, we believe that when a decision by a thyroid nodule biopsy cannot be distinctly made, the combination of our biomarkers may be one of the criteria to be taken into account for the final decision, together with the identification of ANXA1 in serum and saliva.