DNA methylation abnormalities are frequent events in early tumors. Additionally, DNA methylation is relatively stable over time and can be non-invasively detected in blood. Therefore, DNA methylation has a great potential to become an early diagnostic biomarker of cancers. In order to find potential diagnostic markers for lung squamous cell carcinoma (LUSC), a method for identifying LUSC-specific candidate diagnostic markers was proposed. We screened 6 LUSC-specific CpGs by comparing the methylation profiles of 172 samples from LUSC patients, 42 normal lung samples, 184 normal blood samples, and 1 306 samples from patients with other cancers which was collected from TCGA (The Cancer GenomeAtlas) database. A supportvector machine model was constructed to distinguish LUSC patients from normal controls. The combination of six sites achieved 93%-99% sensitivity in predicting LUSC, 100% specificity in excluding all normal samples, and ~ 99% specificity in excluding other cancers. Overall, our study provides promising biomarkers for the diagnosis of LUSC.