Head and neck squamous cell carcinoma (HNSCC) remains one of the most common malignancies associated with poor prognosis. DNA methylation has emerged as an important mechanism underlying the radio-resistance of tumors. Prognostic biomarkers based on radiotherapy-related aberrant DNA methylation are limited. Methylation profiles of 388 patients with HNSCC were acquired from The Cancer Genome Atlas (TCGA) portal. Genes with differentially methylated CpG sites (DMGs) were screened between patients with a favorable and poor prognosis with or without radiotherapy. A weight gene co-methylation network was constructed using a Weighted Gene Co-expression Network Analysis (WGCNA) package. A lasso Cox-PH model was used to identify the optimal panel of genes with the ability to predict survival in these patients. Prognostic performance of the multi-gene methylation signature was assessed in a training set and confirmed in a validation set. A total of 976 DMGs were observed between favorable and poor prognostic samples. Four DMG-enriched co-methylation modules were identified. A four-gene methylation signature was determined by the lasso Cox-PH model that consisted of ZNF10, TMPRSS12, ERGIC2, and RNF215. The risk score based on the four-gene signature was able to divide the training or validation set into two risk groups with significantly different overall survival. Thus, the present study revealed a radiotherapy-related four-gene methylation signature to predict survival outcomes of patients with HNSCC, providing candidate therapeutic targets for novel therapy against HNSCC. However, substantial validation experiments are required.